R Notebook

1 Old 29 Training sets

1.1 Boxplot Top N: None vs Elbow vs Median

top_none <- read.table("~/Box/SCRMshaw_HD_all29Sets_allMethods_Jan2020/pCRMeval/top_noneAmplCurve/output_top_none_oldAllSetsAllMethods.bed",header = TRUE)
top_median_imm <- read.table("~/Box/SCRMshaw_HD_all29Sets_allMethods_Jan2020/pCRMeval/medianAmplitudeCurve/output_topMedianAmp_allold29sets_IMM.bed",header = TRUE)
top_elbow_imm <- read.table("~/Box/SCRMshaw_HD_all29Sets_allMethods_Jan2020/pCRMeval/elbowAmplitudeCurve/output_topAll29_elbow_IMM.bed",header = TRUE)
top_none_imm <-  subset(top_none,top_none$Method=="imm")

  boxplot(top_none_imm$PercentageTrainingSetSensitivity*100,top_median_imm$PercentageTrainingSetSensitivity*100,top_elbow_imm$PercentageTrainingSetSensitivity*100,top_none_imm$PercentageRedflyRecovered*100,top_median_imm$PercentageRedflyRecovered*100,top_elbow_imm$PercentageRedflyRecovered*100,top_none_imm$percentageExpressionPatternPrecision*100,top_median_imm$percentageExpressionPatternPrecision*100,top_elbow_imm$percentageExpressionPatternPrecision*100,names = c("TS_None","TS_Med","TS_Elb","RR_None","RR_Med","RR_Elb","EP_None","EP_Med","EP_Elb"),main="Percentage_comparison_3methods_allmeasures_new74sets",boxwex=0.6,col = c("orange","red","blue","orange","red","blue","orange","red","blue"),ylim=c(0,100))

1.2 Dotplots

1.2.0.1 Training set Sensitivity

library(ggplot2)
#Training set Sensitivity
ggplot(top_none_imm)+
  facet_wrap(~TsetName)+  
  geom_point(data=top_none_imm,aes(x=SCRMs,y=PercentageTrainingSetSensitivity*100,colour="None"),pch=8)+
  geom_point(data=top_median_imm,aes(x=SCRMs,y=PercentageTrainingSetSensitivity*100,colour="Median"),pch=8)+
  geom_point(data=top_elbow_imm,aes(x=SCRMs,y=PercentageTrainingSetSensitivity*100,colour="Elbow"),pch=8)+
  scale_colour_manual(values = c("orange","red","blue","black"))+
  ylim(0,100)+
  theme_bw()

1.2.0.2 Redfly Recovery

#Redfly Recovery
ggplot(top_none_imm)+
  facet_wrap(~TsetName)+  
  geom_point(data=top_none_imm,aes(x=SCRMs,y=PercentageRedflyRecovered*100,colour="None"),pch=8)+
  geom_point(data=top_median_imm,aes(x=SCRMs,y=PercentageRedflyRecovered*100,colour="Median"),pch=8)+
  geom_point(data=top_elbow_imm,aes(x=SCRMs,y=PercentageRedflyRecovered*100,colour="Elbow"),pch=8)+
  scale_colour_manual(values = c("orange","red","blue"))+
  ylim(0,100)+
  theme_bw()

## Warning: Removed 1 rows containing missing values (geom_point).

1.2.0.3 Expression Pattern Precision

#Pattern Precision
ggplot(top_none_imm)+
  facet_wrap(~TsetName)+  
  geom_point(data=top_none_imm,aes(x=SCRMs,y=percentageExpressionPatternPrecision*100,colour="None"),pch=8)+
  geom_point(data=top_median_imm,aes(x=SCRMs,y=percentageExpressionPatternPrecision*100,colour="Median"),pch=8)+
  geom_point(data=top_elbow_imm,aes(x=SCRMs,y=percentageExpressionPatternPrecision*100,colour="Elbow"),pch=8)+
  ylim(0,100)+
  scale_colour_manual(values = c("orange","red","blue"))+
  theme_bw()

1.2.0.4 Expression Pattern Recall

#Pattern Precision permuted comparison
ggplot(top_none_imm)+
  facet_wrap(~TsetName)+  
  geom_point(data=top_none_imm,aes(x=SCRMs,y=percentageExpressionPatternRecall*100,colour="None"),pch=8)+
  geom_point(data=top_median_imm,aes(x=SCRMs,y=ExpectedpercentageExpressionPatternRecall*100,colour="Median"),pch=8)+
  geom_point(data=top_elbow_imm,aes(x=SCRMs,y=ExpectedpercentageExpressionPatternRecall*100,colour="Elbow"),pch=8)+

  ylim(0,100)+
  scale_colour_manual(values = c("orange","red","blue"))+
  theme_bw()

1.3 Old Training set classification

1.3.1 Top Median of Amplitude curve

#random 29
top_fake <-read.table("/Users/hasibaasma/Box/complete_data_representation_newMethod/files/del_randomAvg62sets_28times_Excel_withSpec.bed",header = T)

top_orig_oldSets_Med<- top_median_imm

s1 <- top_orig_oldSets_Med[order(top_orig_oldSets_Med$TsetName),]
s2 <- top_fake[order(top_fake$TsetName),]
subsetIMMevaluationOutputContRand1000_Actual1000_diff<- cbind.data.frame(s1$TsetName,s1$Method,s1$TsetSize,s1$SCRMs,s1$TrainingSetRecovered,s1$PercentageTrainingSetSensitivity,s1$REDflyRecovered,s1$PercentageRedflyRecovered,s1$percentageExpressionPatternRecall-s2$PercentageSpecificity,s2$SCRMs,s2$TrainingSetRecovered,s2$PercentageTrainingSetSensitivity,s2$REDflyRecovered,s2$PercentageRedflyRecovered,s1$PercentageTrainingSetSensitivity - s2$PercentageTrainingSetSensitivity,s1$PercentageRedflyRecovered - s2$PercentageRedflyRecovered)
colnames(subsetIMMevaluationOutputContRand1000_Actual1000_diff)[1] <-"TsetName"
colnames(subsetIMMevaluationOutputContRand1000_Actual1000_diff)[4] <-"SCRMs"
colnames(subsetIMMevaluationOutputContRand1000_Actual1000_diff)[9] <-"DifferenceBetweenActualRandomPercentageSpecificity"
colnames(subsetIMMevaluationOutputContRand1000_Actual1000_diff)[15] <- "DifferenceBetweenActualRandomTsetRecovery"
colnames(subsetIMMevaluationOutputContRand1000_Actual1000_diff)[16] <- "DifferenceBetweenActualRandomRedflyRecovery"

##checking..creating ranges of differences to write in the observations
minimum_differenceBetweenActualRandomRedflyRecovery <- list()
maximum_differenceBetweenActualRandomRedflyRecovery <- list()
median_differenceBetweenActualRandomRedflyRecovery <- list()
minimum_DifferenceBetweenActualTsetRecovery<- list()
maximum_DifferenceBetweenActualTsetRecovery<- list()
median_DifferenceBetweenActualTsetRecovery<- list()
minimum_specificity<- list()
maximum_specificity<- list()
median_specificity<- list()
for(i in unique(subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName)){
  minimum_differenceBetweenActualRandomRedflyRecovery[[i]] <- min(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomRedflyRecovery[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  median_differenceBetweenActualRandomRedflyRecovery[[i]] <- median(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomRedflyRecovery[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  maximum_differenceBetweenActualRandomRedflyRecovery[[i]] <-max(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomRedflyRecovery[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  minimum_DifferenceBetweenActualTsetRecovery[[i]]<-min(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomTsetRecovery[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  maximum_DifferenceBetweenActualTsetRecovery[[i]]<-max(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomTsetRecovery[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  median_DifferenceBetweenActualTsetRecovery[[i]]<-median(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomTsetRecovery[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  
  minimum_specificity[[i]]<-min(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomPercentageSpecificity[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  maximum_specificity[[i]]<-max(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomPercentageSpecificity[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  median_specificity[[i]]<-median(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomPercentageSpecificity[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  
}
library(qdapTools)
minred <- list2df(minimum_differenceBetweenActualRandomRedflyRecovery,"minimum difference to random in redfly recovery","TsetName")
medred <-list2df(median_differenceBetweenActualRandomRedflyRecovery,"median difference to random in redfly recovery","TsetName")
maxred<-list2df(maximum_differenceBetweenActualRandomRedflyRecovery,"maximum difference to random in redfly recovery","TsetName")
minTset<-list2df(minimum_DifferenceBetweenActualTsetRecovery,"minimum difference to random in Tset sensitivity","TsetName")
maxTset<- list2df(maximum_DifferenceBetweenActualTsetRecovery,"maximum difference to random in Tset sensitivity","TsetName")
medTset<-list2df(median_DifferenceBetweenActualTsetRecovery,"median difference to random in Tset sensitivity","TsetName")
minSpec<-list2df(minimum_specificity,"minimum difference to random in specificity","TsetName")
maxSpec<-list2df(maximum_specificity,"maximum difference to random in specificity","TsetName")
medSpec <-list2df(median_specificity,"median difference to random in specificity","TsetName")

df1<- merge(medred,medTset,by="TsetName")
dffinal<-merge(df1,medSpec,by="TsetName")
#dffinal<-merge(df7,by="TsetName")




#data.frame(unclass(table(dffinal)))
#table(dffinal)
write.table(dffinal,file="~/Box/Old_and_newTsets_3postProcMethods_3categories/oldTsets/finaltable_medianAmplitudeCurve_old29tsets.txt",sep = "\t")

library(knitr)
library(kableExtra)
library(dplyr)

## 
## Attaching package: 'dplyr'

## The following object is masked from 'package:qdapTools':
## 
##     id

## The following objects are masked from 'package:stats':
## 
##     filter, lag

## The following objects are masked from 'package:base':
## 
##     intersect, setdiff, setequal, union

1.3.1.1 Good Tset

Conditions being set for these good Tsets include:

• Difference To Random in Redfly Recovery must be greater than or equal to 8percent at cutoff,
• Difference to Random in Tset Sensitivity must be greater than or equalt to 10percent,
• Median of Specificity must be greater than or equalt to 8percent

finaltable<- list()
# 
#setting up conditions Good Tsets OverAll
dffinalDF <- as.data.frame(dffinal)


colnames(dffinalDF) <-c("TsetName","MedianDifferenceToRandomOfRedflyRecovery","MedianDifferenceInTsetSensitivity","MedianDifferenceInTsetSpecificity")


goodTsetOverAll<- filter(dffinalDF,MedianDifferenceToRandomOfRedflyRecovery >=8,  MedianDifferenceInTsetSensitivity >=10,MedianDifferenceInTsetSpecificity >=8)



goodTsetsOverAll <- list()
colnames(goodTsetOverAll) <-c("TsetName","Difference to random in redfly recovery at cutoff","Difference in Tset Sensitivity to random at cutoff","Difference to random in Tset Specificity at cutoff")

goodTsetOverAll<- cbind(goodTsetOverAll[1],goodTsetOverAll[2],goodTsetOverAll[3],goodTsetOverAll[4])
kable_input3<- kable(goodTsetOverAll,digits = 2,caption = "Overall Good Training Sets") 
#column_spec(kable_input3,2:10,width = "1cm")
kable_input3

Overall Good Training Sets

TsetName	Difference to random in redfly recovery at cutoff	Difference in Tset Sensitivity to random at cutoff	Difference to random in Tset Specificity at cutoff

#goodTsetOverAll

1.3.1.2 Good Tsets Ignoring Specificity

Conditions being set for these good Tsets(if we ignore specificity) include:

• Difference To Random in redfly recovery must be greater than or equal to 10percent at cutoff ,
• Difference In Tset Sensitivity to random must be greater than or equalt to 10.

goodTsetsIgnoringSpecificity <- filter(dffinalDF,MedianDifferenceToRandomOfRedflyRecovery >=10,MedianDifferenceInTsetSensitivity>=10)
colnames(goodTsetsIgnoringSpecificity) <-c("TsetName","Difference to random in redfly recovery at cutoff","Difference in Tset Sensitivity to random at cutoff","Difference to random in Tset Specificity at cutoff")
  goodTsetsIgnoringSpecificity <- cbind(goodTsetsIgnoringSpecificity[1],goodTsetsIgnoringSpecificity[2],goodTsetsIgnoringSpecificity[3],goodTsetsIgnoringSpecificity[4])
kable_input4<- kable(goodTsetsIgnoringSpecificity,digits = 2,caption = "Good Training Sets Ignoring Poor Specificity")
#column_spec(kable_input4,2:10,width = "2cm")
kable_input4

Good Training Sets Ignoring Poor Specificity

TsetName	Difference to random in redfly recovery at cutoff	Difference in Tset Sensitivity to random at cutoff	Difference to random in Tset Specificity at cutoff
mapping1.blastoderm	17.59	10.83	5.85
mapping1.glia	12.33	58.39	-1.97
mapping1.malpighian_tubules	10.51	41.72	-7.70
mapping1.tracheal_system	17.34	34.58	-3.49
mapping1.ventral_ectoderm	11.32	36.16	-6.08
mapping1.visceral_mesoderm	23.12	22.67	-5.28
mapping2.glia	17.97	58.39	-1.25

1.3.1.3 Intermediate

These are the sets basically, that do not fall into very good or very bad training sets, mainly because they perform good in two measures but not in the third one. Excluded those sets which were already categorized as good.

Specific conditions being set for these Intermediate sets include:

• Difference To Random in redfly recovery must be grater than 8 and Difference to random in Tset Sensitivity must be greater than 10 but that of specificity could be less than 10, OR
• Difference To Random in redfly recovery must be greater than 7 and specificity must be greater than 5 but Difference to random in Tset Sensitivity could be less than 10 but not less than random, OR
• Difference To random in Tset Sensitivity and specificity must be greater than 10 but Difference to random in redfly recovery could be less than 10 but not less than random.

intermediateTSets <- filter(dffinalDF,MedianDifferenceToRandomOfRedflyRecovery >=8 & MedianDifferenceInTsetSensitivity >=10 &  MedianDifferenceInTsetSpecificity >=0 | MedianDifferenceToRandomOfRedflyRecovery >=0 & MedianDifferenceInTsetSensitivity >=10 &  MedianDifferenceInTsetSpecificity >=5 |MedianDifferenceToRandomOfRedflyRecovery >=7 & MedianDifferenceInTsetSensitivity >=0 &  MedianDifferenceInTsetSpecificity >=5 )
onlyintermediateTSets=data.frame()

onlyintermediateTSetsNum <-    which(!(intermediateTSets$TsetName %in% goodTsetOverAll$TsetName))
  for (i in 1:length(onlyintermediateTSetsNum)){
         onlyintermediateTSetIter <- intermediateTSets[onlyintermediateTSetsNum[i],]
        onlyintermediateTSets <- rbind(onlyintermediateTSets,onlyintermediateTSetIter)       

}



intermediateTSets <- onlyintermediateTSets
colnames(intermediateTSets) <-c("TsetName","Difference to random in redfly recovery at cutoff","Difference in Tset Sensitivity to random at cutoff","Difference to random in Tset Specificity at cutoff")
 intermediateTSets <- cbind(intermediateTSets[1],intermediateTSets[2],intermediateTSets[3],intermediateTSets[4])
kable_input4<- kable(intermediateTSets,digits = 2,caption = "Intermediate Tsets") 
#column_spec(kable_input4,2:10,width = "3cm")
kable_input4

Intermediate Tsets

TsetName	Difference to random in redfly recovery at cutoff	Difference in Tset Sensitivity to random at cutoff	Difference to random in Tset Specificity at cutoff
mapping1.blastoderm	17.59	10.83	5.85

1.3.1.4 Poor Tsets

Conditions being set for these poor sets include. If a set fullfills any of the following condition

• Difference To Random in redfly recovery is less than 0 OR ,
• Difference In Tset Sensitivity is less than or equalt to 0 OR,
• Specificity is less than or equalt to 0,

#poor Tsets

poorTsetOverAll<- filter(dffinalDF, MedianDifferenceToRandomOfRedflyRecovery <=0 |MedianDifferenceInTsetSensitivity <= 0 | MedianDifferenceInTsetSpecificity<=0 )

colnames(poorTsetOverAll) <-c("TsetName","Difference to random in redfly recovery at cutoff","Difference in Tset Sensitivity to random at cutoff","Difference to random in Tset Specificity at cutoff")

poorTsetOverAll<- cbind(poorTsetOverAll[1],poorTsetOverAll[2],poorTsetOverAll[3],poorTsetOverAll[4])
kable_input6<- kable(poorTsetOverAll,digits = 2,caption = "Overall Poor Training Sets")
#column_spec(kable_input6,2:10,width = "2cm")
kable_input6

Overall Poor Training Sets

TsetName	Difference to random in redfly recovery at cutoff	Difference in Tset Sensitivity to random at cutoff	Difference to random in Tset Specificity at cutoff
mapping1.adult_mesoderm	8.53	20.89	-3.92
mapping1.amnioserosa	-0.56	33.39	-8.14
mapping1.cns	0.48	-9.31	-8.15
mapping1.dorsal_ectoderm	6.12	43.00	-5.91
mapping1.ectoderm	9.89	9.33	-6.50
mapping1.endoderm	17.95	3.21	-7.34
mapping1.eye	4.80	33.39	-7.76
mapping1.female_gonad	-8.96	-3.15	-6.64
mapping1.glia	12.33	58.39	-1.97
mapping1.imaginal_disc	6.14	-9.15	-5.96
mapping1.male_gonad	-3.17	5.05	-9.58
mapping1.malpighian_tubules	10.51	41.72	-7.70
mapping1.mesectoderm	7.93	58.39	-6.13
mapping1.mesoderm	0.25	23.77	-5.90
mapping1.neuroectoderm	7.29	36.16	-4.78
mapping1.pns	3.71	6.66	-7.46
mapping1.salivary_gland	23.03	1.24	-5.26
mapping1.somatic_muscle	8.84	5.61	-4.00
mapping1.tracheal_system	17.34	34.58	-3.49
mapping1.ventral_ectoderm	11.32	36.16	-6.08
mapping1.visceral_mesoderm	23.12	22.67	-5.28
mapping2.ectoderm	3.07	0.28	-6.39
mapping2.eye	3.15	11.96	-7.62
mapping2.glia	17.97	58.39	-1.25
mapping2.mesoderm	15.81	-13.61	-6.35
mapping2.neuronal	-3.80	-10.69	-8.32
mapping2.reproductive_system	12.95	-18.28	-9.58
mapping2.wing	8.93	6.01	-5.97

1.3.2 Top Elbow of Amplitude curve

#random 29
top_fake <-read.table("/Users/hasibaasma/Box/complete_data_representation_newMethod/files/del_randomAvg62sets_28times_Excel_withSpec.bed",header = T)

top_orig_oldSets_Med<- top_elbow_imm

s1 <- top_orig_oldSets_Med[order(top_orig_oldSets_Med$TsetName),]
s2 <- top_fake[order(top_fake$TsetName),]
subsetIMMevaluationOutputContRand1000_Actual1000_diff<- cbind.data.frame(s1$TsetName,s1$Method,s1$TsetSize,s1$SCRMs,s1$TrainingSetRecovered,s1$PercentageTrainingSetSensitivity,s1$REDflyRecovered,s1$PercentageRedflyRecovered,s1$percentageExpressionPatternRecall-s2$PercentageSpecificity,s2$SCRMs,s2$TrainingSetRecovered,s2$PercentageTrainingSetSensitivity,s2$REDflyRecovered,s2$PercentageRedflyRecovered,s1$PercentageTrainingSetSensitivity - s2$PercentageTrainingSetSensitivity,s1$PercentageRedflyRecovered - s2$PercentageRedflyRecovered)
colnames(subsetIMMevaluationOutputContRand1000_Actual1000_diff)[1] <-"TsetName"
colnames(subsetIMMevaluationOutputContRand1000_Actual1000_diff)[4] <-"SCRMs"
colnames(subsetIMMevaluationOutputContRand1000_Actual1000_diff)[9] <-"DifferenceBetweenActualRandomPercentageSpecificity"
colnames(subsetIMMevaluationOutputContRand1000_Actual1000_diff)[15] <- "DifferenceBetweenActualRandomTsetRecovery"
colnames(subsetIMMevaluationOutputContRand1000_Actual1000_diff)[16] <- "DifferenceBetweenActualRandomRedflyRecovery"

##checking..creating ranges of differences to write in the observations
minimum_differenceBetweenActualRandomRedflyRecovery <- list()
maximum_differenceBetweenActualRandomRedflyRecovery <- list()
median_differenceBetweenActualRandomRedflyRecovery <- list()
minimum_DifferenceBetweenActualTsetRecovery<- list()
maximum_DifferenceBetweenActualTsetRecovery<- list()
median_DifferenceBetweenActualTsetRecovery<- list()
minimum_specificity<- list()
maximum_specificity<- list()
median_specificity<- list()
for(i in unique(subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName)){
  minimum_differenceBetweenActualRandomRedflyRecovery[[i]] <- min(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomRedflyRecovery[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  median_differenceBetweenActualRandomRedflyRecovery[[i]] <- median(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomRedflyRecovery[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  maximum_differenceBetweenActualRandomRedflyRecovery[[i]] <-max(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomRedflyRecovery[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  minimum_DifferenceBetweenActualTsetRecovery[[i]]<-min(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomTsetRecovery[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  maximum_DifferenceBetweenActualTsetRecovery[[i]]<-max(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomTsetRecovery[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  median_DifferenceBetweenActualTsetRecovery[[i]]<-median(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomTsetRecovery[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  
  minimum_specificity[[i]]<-min(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomPercentageSpecificity[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  maximum_specificity[[i]]<-max(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomPercentageSpecificity[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  median_specificity[[i]]<-median(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomPercentageSpecificity[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  
}
library(qdapTools)
minred <- list2df(minimum_differenceBetweenActualRandomRedflyRecovery,"minimum difference to random in redfly recovery","TsetName")
medred <-list2df(median_differenceBetweenActualRandomRedflyRecovery,"median difference to random in redfly recovery","TsetName")
maxred<-list2df(maximum_differenceBetweenActualRandomRedflyRecovery,"maximum difference to random in redfly recovery","TsetName")
minTset<-list2df(minimum_DifferenceBetweenActualTsetRecovery,"minimum difference to random in Tset sensitivity","TsetName")
maxTset<- list2df(maximum_DifferenceBetweenActualTsetRecovery,"maximum difference to random in Tset sensitivity","TsetName")
medTset<-list2df(median_DifferenceBetweenActualTsetRecovery,"median difference to random in Tset sensitivity","TsetName")
minSpec<-list2df(minimum_specificity,"minimum difference to random in specificity","TsetName")
maxSpec<-list2df(maximum_specificity,"maximum difference to random in specificity","TsetName")
medSpec <-list2df(median_specificity,"median difference to random in specificity","TsetName")

df1<- merge(medred,medTset,by="TsetName")
dffinal<-merge(df1,medSpec,by="TsetName")
#dffinal<-merge(df7,by="TsetName")




#data.frame(unclass(table(dffinal)))
#table(dffinal)
write.table(dffinal,file="~/Box/Old_and_newTsets_3postProcMethods_3categories/oldTsets/finaltable_elbowAmplitudeCurve_old29tsets.txt",sep = "\t")

library(knitr)
library(kableExtra)
library(dplyr)

1.3.2.1 Good Tset

Conditions being set for these good Tsets include:

• Difference To Random in Redfly Recovery must be greater than or equal to 8percent at cutoff,
• Difference to Random in Tset Sensitivity must be greater than or equalt to 10percent,
• Median of Specificity must be greater than or equalt to 8percent

finaltable<- list()
# 
#setting up conditions Good Tsets OverAll
dffinalDF <- as.data.frame(dffinal)


colnames(dffinalDF) <-c("TsetName","MedianDifferenceToRandomOfRedflyRecovery","MedianDifferenceInTsetSensitivity","MedianDifferenceInTsetSpecificity")


goodTsetOverAll<- filter(dffinalDF,MedianDifferenceToRandomOfRedflyRecovery >=8,  MedianDifferenceInTsetSensitivity >=10,MedianDifferenceInTsetSpecificity >=8)



goodTsetsOverAll <- list()
colnames(goodTsetOverAll) <-c("TsetName","Difference to random in redfly recovery at cutoff","Difference in Tset Sensitivity to random at cutoff","Difference to random in Tset Specificity at cutoff")

goodTsetOverAll<- cbind(goodTsetOverAll[1],goodTsetOverAll[2],goodTsetOverAll[3],goodTsetOverAll[4])
kable_input3<- kable(goodTsetOverAll,digits = 2,caption = "Overall Good Training Sets") 
#column_spec(kable_input3,2:10,width = "2cm")

kable_input3

Overall Good Training Sets

TsetName	Difference to random in redfly recovery at cutoff	Difference in Tset Sensitivity to random at cutoff	Difference to random in Tset Specificity at cutoff

1.3.2.2 Good Tsets Ignoring Specificity

Conditions being set for these good Tsets(if we ignore specificity) include:

• Difference To Random in redfly recovery must be greater than or equal to 10percent at cutoff ,
• Difference In Tset Sensitivity to random must be greater than or equalt to 10.

goodTsetsIgnoringSpecificity <- filter(dffinalDF,MedianDifferenceToRandomOfRedflyRecovery >=10,MedianDifferenceInTsetSensitivity>=10)
colnames(goodTsetsIgnoringSpecificity) <-c("TsetName","Difference to random in redfly recovery at cutoff","Difference in Tset Sensitivity to random at cutoff","Difference to random in Tset Specificity at cutoff")
  goodTsetsIgnoringSpecificity <- cbind(goodTsetsIgnoringSpecificity[1],goodTsetsIgnoringSpecificity[2],goodTsetsIgnoringSpecificity[3],goodTsetsIgnoringSpecificity[4])
kable_input4<- kable(goodTsetsIgnoringSpecificity,digits = 2,caption = "Good Training Sets Ignoring Poor Specificity")
#column_spec(kable_input4,2:10,width = "2cm")
kable_input4

Good Training Sets Ignoring Poor Specificity

TsetName	Difference to random in redfly recovery at cutoff	Difference in Tset Sensitivity to random at cutoff	Difference to random in Tset Specificity at cutoff
mapping1.adult_mesoderm	54.97	20.89	-5.81
mapping1.amnioserosa	84.25	33.39	-8.50
mapping1.dorsal_ectoderm	41.76	35.31	-7.89
mapping1.eye	61.63	33.39	-5.95
mapping1.glia	43.14	58.39	-1.97
mapping1.malpighian_tubules	21.82	25.05	-8.64
mapping1.mesectoderm	47.39	58.39	-5.27
mapping1.mesoderm	13.56	12.23	-7.48
mapping1.neuroectoderm	49.97	36.16	-7.98
mapping1.tracheal_system	30.38	29.82	-6.00
mapping1.ventral_ectoderm	26.29	30.61	-7.66
mapping1.visceral_mesoderm	40.18	11.96	-6.23
mapping2.glia	48.66	58.39	-2.29

1.3.2.3 Intermediate

These are the sets basically, that do not fall into very good or very bad training sets, mainly because they perform good in two measures but not in the third one. Excluded those sets which were already categorized as good.

Specific conditions being set for these Intermediate sets include:

• Difference To Random in redfly recovery must be grater than 8 and Difference to random in Tset Sensitivity must be greater than 10 but that of specificity could be less than 10, OR
• Difference To Random in redfly recovery must be greater than 7 and specificity must be greater than 5 but Difference to random in Tset Sensitivity could be less than 10 but not less than random, OR
• Difference To random in Tset Sensitivity and specificity must be greater than 10 but Difference to random in redfly recovery could be less than 10 but not less than random.

intermediateTSets <- filter(dffinalDF,MedianDifferenceToRandomOfRedflyRecovery >=8 & MedianDifferenceInTsetSensitivity >=10 &  MedianDifferenceInTsetSpecificity >=0 | MedianDifferenceToRandomOfRedflyRecovery >=0 & MedianDifferenceInTsetSensitivity >=10 &  MedianDifferenceInTsetSpecificity >=5 |MedianDifferenceToRandomOfRedflyRecovery >=7 & MedianDifferenceInTsetSensitivity >=0 &  MedianDifferenceInTsetSpecificity >=5 )
onlyintermediateTSets=data.frame()

onlyintermediateTSetsNum <-    which(!(intermediateTSets$TsetName %in% goodTsetOverAll$TsetName))
  for (i in 1:length(onlyintermediateTSetsNum)){
         onlyintermediateTSetIter <- intermediateTSets[onlyintermediateTSetsNum[i],]
        onlyintermediateTSets <- rbind(onlyintermediateTSets,onlyintermediateTSetIter)       

}



intermediateTSets <- onlyintermediateTSets
colnames(intermediateTSets) <-c("TsetName","Difference to random in redfly recovery at cutoff","Difference in Tset Sensitivity to random at cutoff","Difference to random in Tset Specificity at cutoff")
 intermediateTSets <- cbind(intermediateTSets[1],intermediateTSets[2],intermediateTSets[3],intermediateTSets[4])
kable_input4<- kable(intermediateTSets,digits = 2,caption = "Intermediate Tsets") 
#column_spec(kable_input4,2:10,width = "3cm")
kable_input4

Intermediate Tsets

	TsetName	Difference to random in redfly recovery at cutoff	Difference in Tset Sensitivity to random at cutoff	Difference to random in Tset Specificity at cutoff
NA	NA	NA	NA	NA

1.3.2.4 Poor Tsets

Conditions being set for these poor sets include. If a set fullfills any of the following condition:

• Difference To Random in redfly recovery is less than 0 OR ,
• Difference In Tset Sensitivity is less than or equalt to 0 OR,
• Specificity is less than or equalt to 0,

#poor Tsets

poorTsetOverAll<- filter(dffinalDF, MedianDifferenceToRandomOfRedflyRecovery <=0 |MedianDifferenceInTsetSensitivity <= 0 | MedianDifferenceInTsetSpecificity<=0 )

colnames(poorTsetOverAll) <-c("TsetName","Difference to random in redfly recovery at cutoff","Difference in Tset Sensitivity to random at cutoff","Difference to random in Tset Specificity at cutoff")

poorTsetOverAll<- cbind(poorTsetOverAll[1],poorTsetOverAll[2],poorTsetOverAll[3],poorTsetOverAll[4])
kable_input6<- kable(poorTsetOverAll,digits = 2,caption = "Overall Poor Training Sets")
#column_spec(kable_input6,2:10,width = "2cm")
kable_input6

Overall Poor Training Sets

TsetName	Difference to random in redfly recovery at cutoff	Difference in Tset Sensitivity to random at cutoff	Difference to random in Tset Specificity at cutoff
mapping1.adult_mesoderm	54.97	20.89	-5.81
mapping1.amnioserosa	84.25	33.39	-8.50
mapping1.cns	0.49	-20.07	-8.82
mapping1.dorsal_ectoderm	41.76	35.31	-7.89
mapping1.ectoderm	10.20	-9.54	-7.92
mapping1.endoderm	51.79	-3.68	-9.17
mapping1.eye	61.63	33.39	-5.95
mapping1.female_gonad	5.26	-10.84	-6.64
mapping1.glia	43.14	58.39	-1.97
mapping1.imaginal_disc	12.33	-16.94	-6.62
mapping1.male_gonad	21.82	5.05	-7.83
mapping1.malpighian_tubules	21.82	25.05	-8.64
mapping1.mesectoderm	47.39	58.39	-5.27
mapping1.mesoderm	13.56	12.23	-7.48
mapping1.neuroectoderm	49.97	36.16	-7.98
mapping1.pns	11.43	-3.68	-7.79
mapping1.salivary_gland	54.58	-5.90	-7.42
mapping1.somatic_muscle	8.89	-2.72	-6.33
mapping1.tracheal_system	30.38	29.82	-6.00
mapping1.ventral_ectoderm	26.29	30.61	-7.66
mapping1.visceral_mesoderm	40.18	11.96	-6.23
mapping2.ectoderm	4.16	-5.13	-7.46
mapping2.eye	19.36	1.24	-8.93
mapping2.glia	48.66	58.39	-2.29
mapping2.mesoderm	28.77	-17.61	-7.18
mapping2.neuronal	8.06	-20.99	-9.17
mapping2.reproductive_system	34.03	-21.61	-8.86
mapping2.wing	9.88	-3.52	-2.96

1.3.3 Top None of Amplitude curve

#random 29
top_fake <-read.table("/Users/hasibaasma/Box/complete_data_representation_newMethod/files/del_randomAvg62sets_28times_Excel_withSpec.bed",header = T)

top_orig_oldSets_Med<- top_none_imm

s1 <- top_orig_oldSets_Med[order(top_orig_oldSets_Med$TsetName),]
s2 <- top_fake[order(top_fake$TsetName),]
subsetIMMevaluationOutputContRand1000_Actual1000_diff<- cbind.data.frame(s1$TsetName,s1$Method,s1$TsetSize,s1$SCRMs,s1$TrainingSetRecovered,s1$PercentageTrainingSetSensitivity,s1$REDflyRecovered,s1$PercentageRedflyRecovered,s1$percentageExpressionPatternRecall-s2$PercentageSpecificity,s2$SCRMs,s2$TrainingSetRecovered,s2$PercentageTrainingSetSensitivity,s2$REDflyRecovered,s2$PercentageRedflyRecovered,s1$PercentageTrainingSetSensitivity - s2$PercentageTrainingSetSensitivity,s1$PercentageRedflyRecovered - s2$PercentageRedflyRecovered)
colnames(subsetIMMevaluationOutputContRand1000_Actual1000_diff)[1] <-"TsetName"
colnames(subsetIMMevaluationOutputContRand1000_Actual1000_diff)[4] <-"SCRMs"
colnames(subsetIMMevaluationOutputContRand1000_Actual1000_diff)[9] <-"DifferenceBetweenActualRandomPercentageSpecificity"
colnames(subsetIMMevaluationOutputContRand1000_Actual1000_diff)[15] <- "DifferenceBetweenActualRandomTsetRecovery"
colnames(subsetIMMevaluationOutputContRand1000_Actual1000_diff)[16] <- "DifferenceBetweenActualRandomRedflyRecovery"

##checking..creating ranges of differences to write in the observations
minimum_differenceBetweenActualRandomRedflyRecovery <- list()
maximum_differenceBetweenActualRandomRedflyRecovery <- list()
median_differenceBetweenActualRandomRedflyRecovery <- list()
minimum_DifferenceBetweenActualTsetRecovery<- list()
maximum_DifferenceBetweenActualTsetRecovery<- list()
median_DifferenceBetweenActualTsetRecovery<- list()
minimum_specificity<- list()
maximum_specificity<- list()
median_specificity<- list()
for(i in unique(subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName)){
  minimum_differenceBetweenActualRandomRedflyRecovery[[i]] <- min(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomRedflyRecovery[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  median_differenceBetweenActualRandomRedflyRecovery[[i]] <- median(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomRedflyRecovery[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  maximum_differenceBetweenActualRandomRedflyRecovery[[i]] <-max(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomRedflyRecovery[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  minimum_DifferenceBetweenActualTsetRecovery[[i]]<-min(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomTsetRecovery[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  maximum_DifferenceBetweenActualTsetRecovery[[i]]<-max(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomTsetRecovery[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  median_DifferenceBetweenActualTsetRecovery[[i]]<-median(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomTsetRecovery[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  
  minimum_specificity[[i]]<-min(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomPercentageSpecificity[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  maximum_specificity[[i]]<-max(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomPercentageSpecificity[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  median_specificity[[i]]<-median(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomPercentageSpecificity[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  
}
library(qdapTools)
minred <- list2df(minimum_differenceBetweenActualRandomRedflyRecovery,"minimum difference to random in redfly recovery","TsetName")
medred <-list2df(median_differenceBetweenActualRandomRedflyRecovery,"median difference to random in redfly recovery","TsetName")
maxred<-list2df(maximum_differenceBetweenActualRandomRedflyRecovery,"maximum difference to random in redfly recovery","TsetName")
minTset<-list2df(minimum_DifferenceBetweenActualTsetRecovery,"minimum difference to random in Tset sensitivity","TsetName")
maxTset<- list2df(maximum_DifferenceBetweenActualTsetRecovery,"maximum difference to random in Tset sensitivity","TsetName")
medTset<-list2df(median_DifferenceBetweenActualTsetRecovery,"median difference to random in Tset sensitivity","TsetName")
minSpec<-list2df(minimum_specificity,"minimum difference to random in specificity","TsetName")
maxSpec<-list2df(maximum_specificity,"maximum difference to random in specificity","TsetName")
medSpec <-list2df(median_specificity,"median difference to random in specificity","TsetName")

df1<- merge(medred,medTset,by="TsetName")
dffinal<-merge(df1,medSpec,by="TsetName")
#dffinal<-merge(df7,by="TsetName")




#data.frame(unclass(table(dffinal)))
#table(dffinal)
write.table(dffinal,file="~/Box/Old_and_newTsets_3postProcMethods_3categories/oldTsets/finaltable_noneAmplitudeCurve_old29tsets.txt",sep = "\t")

library(knitr)
library(kableExtra)
library(dplyr)

1.3.3.1 Good Tset

Conditions being set for these good Tsets include:

• Difference To Random in Redfly Recovery must be greater than or equal to 8percent at cutoff,
• Difference to Random in Tset Sensitivity must be greater than or equalt to 10percent,
• Median of Specificity must be greater than or equalt to 8percent

finaltable<- list()
# 
#setting up conditions Good Tsets OverAll
dffinalDF <- as.data.frame(dffinal)


colnames(dffinalDF) <-c("TsetName","MedianDifferenceToRandomOfRedflyRecovery","MedianDifferenceInTsetSensitivity","MedianDifferenceInTsetSpecificity")


goodTsetOverAll<- filter(dffinalDF,MedianDifferenceToRandomOfRedflyRecovery >=8,  MedianDifferenceInTsetSensitivity >=10,MedianDifferenceInTsetSpecificity >=8)



goodTsetsOverAll <- list()
colnames(goodTsetOverAll) <-c("TsetName","Difference to random in redfly recovery at cutoff","Difference in Tset Sensitivity to random at cutoff","Difference to random in Tset Specificity at cutoff")

goodTsetOverAll<- cbind(goodTsetOverAll[1],goodTsetOverAll[2],goodTsetOverAll[3],goodTsetOverAll[4])
kable_input3<- kable(goodTsetOverAll,digits = 2,caption = "Overall Good Training Sets") 
#column_spec(kable_input3,2:10,width = "2cm")
kable_input3

Overall Good Training Sets

TsetName	Difference to random in redfly recovery at cutoff	Difference in Tset Sensitivity to random at cutoff	Difference to random in Tset Specificity at cutoff

1.3.3.2 Good Tsets Ignoring Specificity

Conditions being set for these good Tsets(if we ignore specificity) include:

• Difference To Random in redfly recovery must be greater than or equal to 10percent at cutoff ,
• Difference In Tset Sensitivity to random must be greater than or equalt to 10.

goodTsetsIgnoringSpecificity <- filter(dffinalDF,MedianDifferenceToRandomOfRedflyRecovery >=10,MedianDifferenceInTsetSensitivity>=10)
colnames(goodTsetsIgnoringSpecificity) <-c("TsetName","Difference to random in redfly recovery at cutoff","Difference in Tset Sensitivity to random at cutoff","Difference to random in Tset Specificity at cutoff")
  goodTsetsIgnoringSpecificity <- cbind(goodTsetsIgnoringSpecificity[1],goodTsetsIgnoringSpecificity[2],goodTsetsIgnoringSpecificity[3],goodTsetsIgnoringSpecificity[4])
kable_input4<- kable(goodTsetsIgnoringSpecificity,digits = 2,caption = "Good Training Sets Ignoring Poor Specificity")
#column_spec(kable_input4,2:10,width = "2cm")
kable_input4

Good Training Sets Ignoring Poor Specificity

TsetName	Difference to random in redfly recovery at cutoff	Difference in Tset Sensitivity to random at cutoff	Difference to random in Tset Specificity at cutoff
mapping1.glia	10.34	58.39	-1.97
mapping1.tracheal_system	13.41	39.34	-1.34
mapping1.visceral_mesoderm	18.26	26.24	-2.88
mapping2.glia	10.09	58.39	-1.25

1.3.3.3 Intermediate

These are the sets basically, that do not fall into very good or very bad training sets, mainly because they perform good in two measures but not in the third one. Excluded those sets which were already categorized as good.

Specific conditions being set for these Intermediate sets include:

• Difference To Random in redfly recovery must be grater than 8 and Difference to random in Tset Sensitivity must be greater than 10 but that of specificity could be less than 10, OR
• Difference To Random in redfly recovery must be greater than 7 and specificity must be greater than 5 but Difference to random in Tset Sensitivity could be less than 10 but not less than random, OR
• Difference To random in Tset Sensitivity and specificity must be greater than 10 but Difference to random in redfly recovery could be less than 10 but not less than random.

intermediateTSets <- filter(dffinalDF,MedianDifferenceToRandomOfRedflyRecovery >=8 & MedianDifferenceInTsetSensitivity >=10 &  MedianDifferenceInTsetSpecificity >=0 | MedianDifferenceToRandomOfRedflyRecovery >=0 & MedianDifferenceInTsetSensitivity >=10 &  MedianDifferenceInTsetSpecificity >=5 |MedianDifferenceToRandomOfRedflyRecovery >=7 & MedianDifferenceInTsetSensitivity >=0 &  MedianDifferenceInTsetSpecificity >=5 )
onlyintermediateTSets=data.frame()

onlyintermediateTSetsNum <-    which(!(intermediateTSets$TsetName %in% goodTsetOverAll$TsetName))
  for (i in 1:length(onlyintermediateTSetsNum)){
         onlyintermediateTSetIter <- intermediateTSets[onlyintermediateTSetsNum[i],]
        onlyintermediateTSets <- rbind(onlyintermediateTSets,onlyintermediateTSetIter)       

}



intermediateTSets <- onlyintermediateTSets
colnames(intermediateTSets) <-c("TsetName","Difference to random in redfly recovery at cutoff","Difference in Tset Sensitivity to random at cutoff","Difference to random in Tset Specificity at cutoff")
 intermediateTSets <- cbind(intermediateTSets[1],intermediateTSets[2],intermediateTSets[3],intermediateTSets[4])
kable_input4<- kable(intermediateTSets,digits = 2,caption = "Intermediate Tsets") 
#column_spec(kable_input4,2:10,width = "3cm")
kable_input4

Intermediate Tsets

TsetName	Difference to random in redfly recovery at cutoff	Difference in Tset Sensitivity to random at cutoff	Difference to random in Tset Specificity at cutoff
mapping1.blastoderm	5.56	18.14	10.17

1.3.3.4 Poor Tsets

Conditions being set for these poor sets include. If a set fullfills any of the following condition

• Difference To Random in redfly recovery is less than 0 OR ,
• Difference In Tset Sensitivity is less than or equalt to 0 OR,
• Specificity is less than or equalt to 0,

#poor Tsets

poorTsetOverAll<- filter(dffinalDF, MedianDifferenceToRandomOfRedflyRecovery <=0 |MedianDifferenceInTsetSensitivity <= 0 | MedianDifferenceInTsetSpecificity<=0 )

colnames(poorTsetOverAll) <-c("TsetName","Difference to random in redfly recovery at cutoff","Difference in Tset Sensitivity to random at cutoff","Difference to random in Tset Specificity at cutoff")

poorTsetOverAll<- cbind(poorTsetOverAll[1],poorTsetOverAll[2],poorTsetOverAll[3],poorTsetOverAll[4])
kable_input6<- kable(poorTsetOverAll,digits = 2,caption = "Overall Poor Training Sets")
#column_spec(kable_input6,2:10,width = "2cm")
kable_input6

Overall Poor Training Sets

TsetName	Difference to random in redfly recovery at cutoff	Difference in Tset Sensitivity to random at cutoff	Difference to random in Tset Specificity at cutoff
mapping1.adult_mesoderm	3.86	20.89	-3.92
mapping1.amnioserosa	-0.65	33.39	-8.14
mapping1.cns	1.17	1.46	-6.46
mapping1.dorsal_ectoderm	3.59	50.69	-5.35
mapping1.ectoderm	6.08	14.99	-4.75
mapping1.endoderm	11.88	6.66	-6.25
mapping1.eye	-0.22	33.39	-7.76
mapping1.female_gonad	-11.71	4.54	-5.17
mapping1.glia	10.34	58.39	-1.97
mapping1.imaginal_disc	6.70	10.34	-4.65
mapping1.male_gonad	-6.34	5.05	-7.83
mapping1.malpighian_tubules	5.54	41.72	-7.70
mapping1.mesectoderm	9.94	58.39	-5.27
mapping1.mesoderm	-0.24	31.46	-3.79
mapping1.neuroectoderm	1.03	36.16	-4.78
mapping1.pns	-0.35	15.28	-6.15
mapping1.salivary_gland	18.50	1.24	-3.10
mapping1.somatic_muscle	8.27	8.39	-0.75
mapping1.tracheal_system	13.41	39.34	-1.34
mapping1.ventral_ectoderm	6.77	36.16	-5.03
mapping1.visceral_mesoderm	18.26	26.24	-2.88
mapping2.ectoderm	0.46	11.09	-4.03
mapping2.eye	0.29	11.96	-6.97
mapping2.glia	10.09	58.39	-1.25
mapping2.mesoderm	11.28	-4.61	-4.04
mapping2.neuronal	-2.12	-4.50	-7.30
mapping2.reproductive_system	10.57	-11.61	-8.86
mapping2.wing	7.88	20.29	-2.96

2 New 72 Training sets

New 74 training sets constructed on January 2020

2.1 Boxplot New Tsets: None vs Elbow vs Median

top_none <- read.table("~/Box/NewTrainingSetsJan2020/pCRMeval_June20_postProcM/None/output_topNone_allNew74sets_IMM.bed",header = TRUE)

top_median_imm <- read.table("~/Box/NewTrainingSetsJan2020/pCRMeval_June20_postProcM/Median/output_topMed_allNew74sets_IMM.bed",header = TRUE)
#top_median_imm <- read.table("~/Box/output_topMedianAmp_allold29sets_IMM copy.bed",header = TRUE)
top_elbow_imm <- read.table("~/Box/NewTrainingSetsJan2020/pCRMeval_June20_postProcM/elbow/output_topElbow_allNew74sets_IMM.bed",header = TRUE)
#top_elbow_imm <- read.table("~/Box/output_topAll29_elbow copy.bed",header = TRUE)
top_none_imm <-  subset(top_none,top_none$Method=="imm")
  boxplot(top_none_imm$PercentageTrainingSetSensitivity*100,top_median_imm$PercentageTrainingSetSensitivity*100,top_elbow_imm$PercentageTrainingSetSensitivity*100,top_none_imm$PercentageRedflyRecovered*100,top_median_imm$PercentageRedflyRecovered*100,top_elbow_imm$PercentageRedflyRecovered*100,top_none_imm$percentageExpressionPatternPrecision*100,top_median_imm$percentageExpressionPatternPrecision*100,top_elbow_imm$percentageExpressionPatternPrecision*100,names = c("TS_None","TS_Med","TS_Elb","RR_None","RR_Med","RR_Elb","EP_None","EP_Med","EP_Elb"),main="Percentage_comparison_3methods_allmeasures_new74sets",boxwex=0.6,col = c("orange","red","blue","orange","red","blue","orange","red","blue"),ylim=c(0,100))

2.2 Dotplots 3 top N methods

2.2.0.1 Training set Sensitivity

library(ggplot2)
#Training set Sensitivity
ggplot(top_none_imm)+
  facet_wrap(~TsetName)+  
  geom_point(data=top_none_imm,aes(x=SCRMs,y=PercentageTrainingSetSensitivity*100,colour="None"),pch=8)+
  geom_point(data=top_median_imm,aes(x=SCRMs,y=PercentageTrainingSetSensitivity*100,colour="Median"),pch=8)+
  geom_point(data=top_elbow_imm,aes(x=SCRMs,y=PercentageTrainingSetSensitivity*100,colour="Elbow"),pch=8)+
  scale_colour_manual(values = c("orange","red","blue","black"))+
  ylim(0,100)+
  theme_bw()

2.2.0.2 Redfly Recovery

#Redfly Recovery
ggplot(top_none_imm)+
  facet_wrap(~TsetName)+  
  geom_point(data=top_none_imm,aes(x=SCRMs,y=PercentageRedflyRecovered*100,colour="None"),pch=8)+
  geom_point(data=top_median_imm,aes(x=SCRMs,y=PercentageRedflyRecovered*100,colour="Median"),pch=8)+
  geom_point(data=top_elbow_imm,aes(x=SCRMs,y=PercentageRedflyRecovered*100,colour="Elbow"),pch=8)+
  scale_colour_manual(values = c("orange","red","blue"))+
  ylim(0,100)+
  theme_bw()

2.2.0.3 Expression Pattern Precision

#Pattern Precision
ggplot(top_none_imm)+
  facet_wrap(~TsetName)+  
  geom_point(data=top_none_imm,aes(x=SCRMs,y=percentageExpressionPatternPrecision*100,colour="None"),pch=8)+
  geom_point(data=top_median_imm,aes(x=SCRMs,y=percentageExpressionPatternPrecision*100,colour="Median"),pch=8)+
  geom_point(data=top_elbow_imm,aes(x=SCRMs,y=percentageExpressionPatternPrecision*100,colour="Elbow"),pch=8)+
  ylim(0,100)+
  scale_colour_manual(values = c("orange","red","blue"))+
  theme_bw()

2.2.0.4 Expression Pattern Recall

#Pattern Precision permuted comparison
ggplot(top_none_imm)+
  facet_wrap(~TsetName)+  
  geom_point(data=top_none_imm,aes(x=SCRMs,y=percentageExpressionPatternRecall*100,colour="None"),pch=8)+
  geom_point(data=top_median_imm,aes(x=SCRMs,y=ExpectedpercentageExpressionPatternRecall*100,colour="Median"),pch=8)+
  geom_point(data=top_elbow_imm,aes(x=SCRMs,y=ExpectedpercentageExpressionPatternRecall*100,colour="Elbow"),pch=8)+

  ylim(0,100)+
  scale_colour_manual(values = c("orange","red","blue"))+
  theme_bw()

2.3 New Training sets classification

2.3.1 Top Median of Amplitude curve

#random 29
top_fake <-read.table("/Users/hasibaasma/Box/NewTrainingSetsJan2020/R/random62oldRun/old/del_randomMEDIAN62sets_75times_Excel_withSpec.txt",header = T)

top_orig_newMed<- top_median_imm

s1 <- top_orig_newMed[order(top_orig_newMed$TsetName),]
s2 <- top_fake[order(top_fake$TsetName),]
subsetIMMevaluationOutputContRand1000_Actual1000_diff<- cbind.data.frame(s1$TsetName,s1$Method,s1$TsetSize,s1$SCRMs,s1$TrainingSetRecovered,s1$PercentageTrainingSetSensitivity,s1$REDflyRecovered,s1$PercentageRedflyRecovered,s1$percentageExpressionPatternRecall-s1$ExpectedpercentageExpressionPatternRecall,s2$SCRMs,s2$TrainingSetRecovered,s2$PercentageTrainingSetSensitivity,s2$REDflyRecovered,s2$PercentageRedflyRecovered,s1$PercentageTrainingSetSensitivity - s2$PercentageTrainingSetSensitivity,s1$PercentageRedflyRecovered - s2$PercentageRedflyRecovered)
colnames(subsetIMMevaluationOutputContRand1000_Actual1000_diff)[1] <-"TsetName"
colnames(subsetIMMevaluationOutputContRand1000_Actual1000_diff)[4] <-"SCRMs"
colnames(subsetIMMevaluationOutputContRand1000_Actual1000_diff)[9] <-"DifferenceBetweenActualRandomPercentageSpecificity"
colnames(subsetIMMevaluationOutputContRand1000_Actual1000_diff)[15] <- "DifferenceBetweenActualRandomTsetRecovery"
colnames(subsetIMMevaluationOutputContRand1000_Actual1000_diff)[16] <- "DifferenceBetweenActualRandomRedflyRecovery"

##checking..creating ranges of differences to write in the observations
minimum_differenceBetweenActualRandomRedflyRecovery <- list()
maximum_differenceBetweenActualRandomRedflyRecovery <- list()
median_differenceBetweenActualRandomRedflyRecovery <- list()
minimum_DifferenceBetweenActualTsetRecovery<- list()
maximum_DifferenceBetweenActualTsetRecovery<- list()
median_DifferenceBetweenActualTsetRecovery<- list()
minimum_specificity<- list()
maximum_specificity<- list()
median_specificity<- list()
for(i in unique(subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName)){
  minimum_differenceBetweenActualRandomRedflyRecovery[[i]] <- min(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomRedflyRecovery[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  median_differenceBetweenActualRandomRedflyRecovery[[i]] <- median(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomRedflyRecovery[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  maximum_differenceBetweenActualRandomRedflyRecovery[[i]] <-max(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomRedflyRecovery[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  minimum_DifferenceBetweenActualTsetRecovery[[i]]<-min(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomTsetRecovery[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  maximum_DifferenceBetweenActualTsetRecovery[[i]]<-max(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomTsetRecovery[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  median_DifferenceBetweenActualTsetRecovery[[i]]<-median(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomTsetRecovery[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  
  minimum_specificity[[i]]<-min(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomPercentageSpecificity[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  maximum_specificity[[i]]<-max(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomPercentageSpecificity[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  median_specificity[[i]]<-median(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomPercentageSpecificity[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  
}
library(qdapTools)
minred <- list2df(minimum_differenceBetweenActualRandomRedflyRecovery,"minimum difference to random in redfly recovery","TsetName")
medred <-list2df(median_differenceBetweenActualRandomRedflyRecovery,"median difference to random in redfly recovery","TsetName")
maxred<-list2df(maximum_differenceBetweenActualRandomRedflyRecovery,"maximum difference to random in redfly recovery","TsetName")
minTset<-list2df(minimum_DifferenceBetweenActualTsetRecovery,"minimum difference to random in Tset sensitivity","TsetName")
maxTset<- list2df(maximum_DifferenceBetweenActualTsetRecovery,"maximum difference to random in Tset sensitivity","TsetName")
medTset<-list2df(median_DifferenceBetweenActualTsetRecovery,"median difference to random in Tset sensitivity","TsetName")
minSpec<-list2df(minimum_specificity,"minimum difference to random in specificity","TsetName")
maxSpec<-list2df(maximum_specificity,"maximum difference to random in specificity","TsetName")
medSpec <-list2df(median_specificity,"median difference to random in specificity","TsetName")

df1<- merge(medred,medTset,by="TsetName")
dffinal<-merge(df1,medSpec,by="TsetName")
#dffinal<-merge(df7,by="TsetName")




#data.frame(unclass(table(dffinal)))
#table(dffinal)
write.table(dffinal,file="~/Box/Old_and_newTsets_3postProcMethods_3categories/newTsets/finaltable_new74sets_medianAmplitudeCurve.txt",sep = "\t")

library(knitr)
library(kableExtra)
library(dplyr)

2.3.1.1 Good Tsets

Conditions being set for these good Tsets include:

• Difference To Random in Redfly Recovery must be greater than or equal to 8percent at cutoff,
• Difference to Random in Tset Sensitivity must be greater than or equalt to 8percent,
• Difference to Expected in Specificity must be greater than or equalt to 4percent

finaltable<- list()
# 
#setting up conditions Good Tsets OverAll
dffinalDF <- as.data.frame(dffinal)


colnames(dffinalDF) <-c("TsetName","MedianDifferenceToRandomOfRedflyRecovery","MedianDifferenceInTsetSensitivity","MedianDifferenceToPermutedInTsetSpecificity")


goodTsetOverAll<- filter(dffinalDF,MedianDifferenceToRandomOfRedflyRecovery >=8,  MedianDifferenceInTsetSensitivity >=8,MedianDifferenceToPermutedInTsetSpecificity >=4)



goodTsetsOverAll <- list()
colnames(goodTsetOverAll) <-c("TsetName","Difference to random in redfly recovery at cutoff","Difference in Tset Sensitivity to random at cutoff","Difference to permuted in Tset Specificity at cutoff")

goodTsetOverAll<- cbind(goodTsetOverAll[1],goodTsetOverAll[2],goodTsetOverAll[3],goodTsetOverAll[4])
kable_input3<- kable(goodTsetOverAll,digits = 2,caption = "Overall Good Training Sets with TS and RR > 8 and SP > 4") 
#column_spec(kable_input3,2:10,width = "2cm")
kable_input3

Overall Good Training Sets with TS and RR > 8 and SP > 4

TsetName	Difference to random in redfly recovery at cutoff	Difference in Tset Sensitivity to random at cutoff	Difference to permuted in Tset Specificity at cutoff
adult_mesoderm.mapping1	32.98	23.33	8.83
adult_muscle	24.87	35.00	10.89
adult_somatic_muscle	27.58	20.00	12.20
blastoderm.mapping1	25.30	32.92	11.09
embryonic_muscle	31.37	21.29	4.17
eye.mapping1	30.36	30.00	5.30
mesectoderm.mapping1	23.68	25.00	4.51

#write.table(goodTsetOverAll,file="~/Box/NewTrainingSetsJan2020/R/goodTsetOverAll.txt",sep = "\t")

2.3.1.2 Good Tsets Ignoring Specificity

Conditions being set for these good Tsets(if we ignore specificity) include:

• Difference To Random in redfly recovery must be greater than or equal to 10percent at cutoff,
• Difference In Tset Sensitivity to random must be greater than or equalt to 10.

goodTsetsIgnoringSpecificity <- filter(dffinalDF,MedianDifferenceToRandomOfRedflyRecovery >=10,MedianDifferenceInTsetSensitivity>=10)
colnames(goodTsetsIgnoringSpecificity) <-c("TsetName","Difference to random in redfly recovery at cutoff","Difference in Tset Sensitivity to random at cutoff","Difference to permuted in Tset Specificity at cutoff")
goodTsetsIgnoringSpecificity <- cbind(goodTsetsIgnoringSpecificity[1],goodTsetsIgnoringSpecificity[2],goodTsetsIgnoringSpecificity[3],goodTsetsIgnoringSpecificity[4])
kable_input4<- kable(goodTsetsIgnoringSpecificity,digits = 2,caption = "Good Training Sets Ignoring Poor Specificity both > 10")
#column_spec(kable_input4,2:10,width = "2cm")
kable_input4

Good Training Sets Ignoring Poor Specificity both > 10

TsetName	Difference to random in redfly recovery at cutoff	Difference in Tset Sensitivity to random at cutoff	Difference to permuted in Tset Specificity at cutoff
adult_mesoderm.mapping1	32.98	23.33	8.83
adult_muscle	24.87	35.00	10.89
adult_somatic_muscle	27.58	20.00	12.20
blastoderm.mapping1	25.30	32.92	11.09
cardiac.mapping1	32.50	13.57	3.30
dorsal_ectoderm.mapping1	20.73	19.37	2.16
emb-larv_circulatory_system	27.42	16.00	2.20
emb-larv_hindgut	24.66	45.71	2.70
emb-larv_visceral	23.90	15.56	0.78
embryonic_midgut	31.79	12.94	2.34
embryonic_muscle	31.37	21.29	4.17
embryonic_somatic_muscle	26.82	32.00	3.21
embryonic_trachea	21.54	22.50	3.33
eye.mapping1	30.36	30.00	5.30
fat_body.mapping1	12.75	28.75	2.59
haltere_disc	29.99	15.56	0.64
mesectoderm.mapping1	23.68	25.00	4.51
mesoderm.mapping1	28.11	17.78	3.25
myoblast	20.77	26.67	1.35
ventral_ectoderm.mapping1	15.97	21.90	1.44

#write.table(goodTsetsIgnoringSpecificity,file="~/Box/NewTrainingSetsJan2020/R/goodTsetsIgnoringSpecificity.txt",sep = "\t")

2.3.1.3 Intermediate Tsets

These are the sets basically, that do not fall into very good or very bad training sets, mainly because they perform good in two measures but not in the third one. Not anymore , Excluded those sets which were already categorized as good.

Specific conditions being set for these Intermediate sets include:

• Difference To Random in redfly recovery must be grater than 8 and Difference to random in Tset Sensitivity must be greater than 8 but that of specificity should just be greater than 0 , OR
• Difference To Random in redfly recovery must be greater than 8 and specificity must be greater than 4 but Difference to random in Tset Sensitivity should just be greater than 0 , OR
• Difference To random in Tset Sensitivity and specificity must be greater than 8 and 4 respectively but Difference to random in redfly recovery should just be greater than 0.

intermediateTSets <- filter(dffinalDF,MedianDifferenceToRandomOfRedflyRecovery >=8 & MedianDifferenceInTsetSensitivity >=8 &  MedianDifferenceToPermutedInTsetSpecificity >=0 | MedianDifferenceToRandomOfRedflyRecovery >=0 & MedianDifferenceInTsetSensitivity >=8 &  MedianDifferenceToPermutedInTsetSpecificity >=4 |MedianDifferenceToRandomOfRedflyRecovery >=8 & MedianDifferenceInTsetSensitivity >=0 &  MedianDifferenceToPermutedInTsetSpecificity >=4 )
onlyintermediateTSets=data.frame()

# onlyintermediateTSetsNum <-    which(!(intermediateTSets$TsetName %in% goodTsetOverAll$TsetName))
# for (i in 1:length(onlyintermediateTSetsNum)){
# onlyintermediateTSetIter <- intermediateTSets[onlyintermediateTSetsNum[i],]
# onlyintermediateTSets <- rbind(onlyintermediateTSets,onlyintermediateTSetIter)       
# 
# }



#intermediateTSets <- onlyintermediateTSets
colnames(intermediateTSets) <-c("TsetName","Difference to random in redfly recovery at cutoff","Difference in Tset Sensitivity to random at cutoff","Difference to random in Tset Specificity at cutoff")
intermediateTSets <- cbind(intermediateTSets[1],intermediateTSets[2],intermediateTSets[3],intermediateTSets[4])
kable_input41<- kable(intermediateTSets,digits = 2,caption = "Intermediate Tsets with all 3 interm") 
#column_spec(kable_input41,2:10,width = "3cm")
kable_input41

Intermediate Tsets with all 3 interm

TsetName	Difference to random in redfly recovery at cutoff	Difference in Tset Sensitivity to random at cutoff	Difference to random in Tset Specificity at cutoff
adult_mesoderm.mapping1	32.98	23.33	8.83
adult_muscle	24.87	35.00	10.89
adult_pns	0.94	20.00	5.06
adult_somatic_muscle	27.58	20.00	12.20
amnioserosa.mapping1	31.40	7.37	5.41
antennal_lobe	20.40	8.39	0.99
blastoderm.mapping1	25.30	32.92	11.09
cardiac.mapping1	32.50	13.57	3.30
dorsal_ectoderm.mapping1	20.73	19.37	2.16
emb-larv_circulatory_system	27.42	16.00	2.20
emb-larv_excretory	31.83	10.00	1.96
emb-larv_foregut	22.92	4.44	4.91
emb-larv_hindgut	24.66	45.71	2.70
emb-larv_visceral	23.90	15.56	0.78
embryonic_epidermis	37.52	6.67	4.70
embryonic_midgut	31.79	12.94	2.34
embryonic_muscle	31.37	21.29	4.17
embryonic_pns	26.26	10.00	2.49
embryonic_sense_organ	1.62	26.67	8.16
embryonic_somatic_muscle	26.82	32.00	3.21
embryonic_trachea	21.54	22.50	3.33
eye.mapping1	30.36	30.00	5.30
fat_body.mapping1	12.75	28.75	2.59
haltere_disc	29.99	15.56	0.64
leg_disc	21.28	10.00	1.76
malpig.mapping1	33.72	10.00	2.85
mesectoderm.mapping1	23.68	25.00	4.51
mesoderm.mapping1	28.11	17.78	3.25
myoblast	20.77	26.67	1.35
salivary.mapping1	26.37	2.11	5.81
ventral_ectoderm.mapping1	15.97	21.90	1.44

#write.table(intermediateTSets,file="~/Box/NewTrainingSetsJan2020/R/intermediateTSets.txt",sep = "\t")

2.3.1.4 Poor Tsets

Conditions being set for these poor sets include. If a set is poor in two categories then its a poor set

• Difference To Random in redfly recovery is less than or equalt to 8 and Difference To Random in Tset Sensitivity is also less than or equalt to 8 OR ,
• Difference To Random in Tset Sensitivity is less than or equalt to 8 and Difference To Permuted in Tset Specificity is also less than or equalt to 4 OR,
• Difference Random in redfly recovery is less than or equalt to 8 and Difference To Permuted in Tset Specificity is also less than or equalt to 4 ,

#poor Tsets

poorTsetOverAll<- filter(dffinalDF, MedianDifferenceToRandomOfRedflyRecovery <=8 & MedianDifferenceInTsetSensitivity <= 8 | MedianDifferenceInTsetSensitivity <= 8 & MedianDifferenceToPermutedInTsetSpecificity<=4 |MedianDifferenceToPermutedInTsetSpecificity<=4 & MedianDifferenceToRandomOfRedflyRecovery <=8 )

colnames(poorTsetOverAll) <-c("TsetName","Difference to random in redfly recovery at cutoff","Difference in Tset Sensitivity to random at cutoff","Difference to random in Tset Specificity at cutoff")

poorTsetOverAll<- cbind(poorTsetOverAll[1],poorTsetOverAll[2],poorTsetOverAll[3],poorTsetOverAll[4])
kable_input6<- kable(poorTsetOverAll,digits = 2,caption = "Overall Poor Training Sets with any one below 0")
#column_spec(kable_input6,2:10,width = "2cm")
kable_input6

Overall Poor Training Sets with any one below 0

TsetName	Difference to random in redfly recovery at cutoff	Difference in Tset Sensitivity to random at cutoff	Difference to random in Tset Specificity at cutoff
adult_brain	12.23	-30.91	1.06
adult_circulatory	24.60	5.00	2.73
adult_cns	12.68	-32.68	1.18
adult_midgut	22.27	7.37	-0.31
adult_nervous	16.47	-35.05	1.33
adult_sense_organ	9.52	3.75	2.76
antenna	18.67	-30.00	0.96
antennal_disc	7.86	-1.90	0.74
disc.mapping1	22.36	-6.67	2.24
disc.mapping2	22.23	3.24	1.98
ectoderm.mapping1	24.18	-12.41	1.47
ectoderm.mapping2	26.67	-21.11	1.08
emb-larv_fat_body	7.57	35.00	2.78
emb-larval_cns	0.23	-4.29	0.35
emb-larval_mushroombody	15.23	2.50	1.72
emb-larval_neuron	23.51	-31.78	0.30
emb-larval_opticlobe	29.35	4.83	1.84
embryonic_salivary	1.91	31.43	2.29
embryonic_ventral_nervous_system	27.27	-17.94	1.63
endoderm.mapping1	29.93	-5.38	2.02
eye_disc	18.57	2.86	2.13
eye-antennal_disc	22.21	-2.22	3.26
eye.mapping2	30.94	-19.69	2.02
female_reproductive	18.67	8.00	2.94
genital_disc	28.00	-4.29	1.40
glia	24.81	-3.64	1.40
glia.mapping1	21.17	2.50	2.00
glia.mapping2	22.68	0.48	2.19
gonad	30.24	-34.74	3.09
imaginal_disc	20.44	-16.06	1.94
leg	25.76	0.00	1.82
mesoderm.mapping2	35.07	-31.10	2.16
neuron	23.04	-34.81	0.52
pns.mapping1	22.78	5.28	2.01
reproductive.mapping2	29.73	-35.65	3.33
somatic_muscle.mapping1	29.34	7.83	3.57
trachea.mapping1	21.62	1.67	1.55
visceral.mapping1	27.43	6.34	2.78
wing.mapping2	22.02	-0.61	2.02

#write.table(poorTsetOverAll,file="~/Box/NewTrainingSetsJan2020/R/poorTsetOverAll.txt",sep = "\t")

2.3.2 Top Elbow of Amplitude curve

#random 29
top_fake <-read.table("/Users/hasibaasma/Box/NewTrainingSetsJan2020/R/random62oldRun/old/del_randomMEDIAN62sets_75times_Excel_withSpec.txt",header = T)

top_orig_newMed<- top_elbow_imm

s1 <- top_orig_newMed[order(top_orig_newMed$TsetName),]
s2 <- top_fake[order(top_fake$TsetName),]
subsetIMMevaluationOutputContRand1000_Actual1000_diff<- cbind.data.frame(s1$TsetName,s1$Method,s1$TsetSize,s1$SCRMs,s1$TrainingSetRecovered,s1$PercentageTrainingSetSensitivity,s1$REDflyRecovered,s1$PercentageRedflyRecovered,s1$percentageExpressionPatternRecall-s1$ExpectedpercentageExpressionPatternRecall,s2$SCRMs,s2$TrainingSetRecovered,s2$PercentageTrainingSetSensitivity,s2$REDflyRecovered,s2$PercentageRedflyRecovered,s1$PercentageTrainingSetSensitivity - s2$PercentageTrainingSetSensitivity,s1$PercentageRedflyRecovered - s2$PercentageRedflyRecovered)
colnames(subsetIMMevaluationOutputContRand1000_Actual1000_diff)[1] <-"TsetName"
colnames(subsetIMMevaluationOutputContRand1000_Actual1000_diff)[4] <-"SCRMs"
colnames(subsetIMMevaluationOutputContRand1000_Actual1000_diff)[9] <-"DifferenceBetweenActualRandomPercentageSpecificity"
colnames(subsetIMMevaluationOutputContRand1000_Actual1000_diff)[15] <- "DifferenceBetweenActualRandomTsetRecovery"
colnames(subsetIMMevaluationOutputContRand1000_Actual1000_diff)[16] <- "DifferenceBetweenActualRandomRedflyRecovery"

##checking..creating ranges of differences to write in the observations
minimum_differenceBetweenActualRandomRedflyRecovery <- list()
maximum_differenceBetweenActualRandomRedflyRecovery <- list()
median_differenceBetweenActualRandomRedflyRecovery <- list()
minimum_DifferenceBetweenActualTsetRecovery<- list()
maximum_DifferenceBetweenActualTsetRecovery<- list()
median_DifferenceBetweenActualTsetRecovery<- list()
minimum_specificity<- list()
maximum_specificity<- list()
median_specificity<- list()
for(i in unique(subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName)){
  minimum_differenceBetweenActualRandomRedflyRecovery[[i]] <- min(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomRedflyRecovery[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  median_differenceBetweenActualRandomRedflyRecovery[[i]] <- median(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomRedflyRecovery[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  maximum_differenceBetweenActualRandomRedflyRecovery[[i]] <-max(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomRedflyRecovery[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  minimum_DifferenceBetweenActualTsetRecovery[[i]]<-min(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomTsetRecovery[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  maximum_DifferenceBetweenActualTsetRecovery[[i]]<-max(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomTsetRecovery[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  median_DifferenceBetweenActualTsetRecovery[[i]]<-median(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomTsetRecovery[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  
  minimum_specificity[[i]]<-min(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomPercentageSpecificity[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  maximum_specificity[[i]]<-max(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomPercentageSpecificity[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  median_specificity[[i]]<-median(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomPercentageSpecificity[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  
}
library(qdapTools)
minred <- list2df(minimum_differenceBetweenActualRandomRedflyRecovery,"minimum difference to random in redfly recovery","TsetName")
medred <-list2df(median_differenceBetweenActualRandomRedflyRecovery,"median difference to random in redfly recovery","TsetName")
maxred<-list2df(maximum_differenceBetweenActualRandomRedflyRecovery,"maximum difference to random in redfly recovery","TsetName")
minTset<-list2df(minimum_DifferenceBetweenActualTsetRecovery,"minimum difference to random in Tset sensitivity","TsetName")
maxTset<- list2df(maximum_DifferenceBetweenActualTsetRecovery,"maximum difference to random in Tset sensitivity","TsetName")
medTset<-list2df(median_DifferenceBetweenActualTsetRecovery,"median difference to random in Tset sensitivity","TsetName")
minSpec<-list2df(minimum_specificity,"minimum difference to random in specificity","TsetName")
maxSpec<-list2df(maximum_specificity,"maximum difference to random in specificity","TsetName")
medSpec <-list2df(median_specificity,"median difference to random in specificity","TsetName")

df1<- merge(medred,medTset,by="TsetName")
dffinal<-merge(df1,medSpec,by="TsetName")
#dffinal<-merge(df7,by="TsetName")




#data.frame(unclass(table(dffinal)))
#table(dffinal)
write.table(dffinal,file="~/Box/Old_and_newTsets_3postProcMethods_3categories/newTsets/finaltable_new74sets_elbowAmplitudeCurve.txt",sep = "\t")

library(knitr)
library(kableExtra)
library(dplyr)

2.3.2.1 Good Tsets

Conditions being set for these good Tsets include:

• Difference To Random in Redfly Recovery must be greater than or equal to 8percent at cutoff,
• Difference to Random in Tset Sensitivity must be greater than or equalt to 8percent,
• Difference to Expected in Specificity must be greater than or equalt to 4percent

finaltable<- list()
# 
#setting up conditions Good Tsets OverAll
dffinalDF <- as.data.frame(dffinal)


colnames(dffinalDF) <-c("TsetName","MedianDifferenceToRandomOfRedflyRecovery","MedianDifferenceInTsetSensitivity","MedianDifferenceToPermutedInTsetSpecificity")


goodTsetOverAll<- filter(dffinalDF,MedianDifferenceToRandomOfRedflyRecovery >=8,  MedianDifferenceInTsetSensitivity >=8,MedianDifferenceToPermutedInTsetSpecificity >=4)



goodTsetsOverAll <- list()
colnames(goodTsetOverAll) <-c("TsetName","Difference to random in redfly recovery at cutoff","Difference in Tset Sensitivity to random at cutoff","Difference to permuted in Tset Specificity at cutoff")

goodTsetOverAll<- cbind(goodTsetOverAll[1],goodTsetOverAll[2],goodTsetOverAll[3],goodTsetOverAll[4])
kable_input3<- kable(goodTsetOverAll,digits = 2,caption = "Overall Good Training Sets with TS and RR > 8 and SP > 4") 
#column_spec(kable_input3,2:10,width = "2cm")
kable_input3

Overall Good Training Sets with TS and RR > 8 and SP > 4

TsetName	Difference to random in redfly recovery at cutoff	Difference in Tset Sensitivity to random at cutoff	Difference to permuted in Tset Specificity at cutoff
adult_muscle	38.49	28.75	11.11
adult_somatic_muscle	41.32	20.00	12.20
blastoderm.mapping1	47.60	22.50	6.62

#write.table(goodTsetOverAll,file="~/Box/NewTrainingSetsJan2020/R/goodTsetOverAll.txt",sep = "\t")

2.3.2.2 Good Tsets Ignoring Specificity

Conditions being set for these good Tsets(if we ignore specificity) include:

• Difference To Random in redfly recovery must be greater than or equal to 10percent at cutoff,
• Difference In Tset Sensitivity to random must be greater than or equalt to 10.

goodTsetsIgnoringSpecificity <- filter(dffinalDF,MedianDifferenceToRandomOfRedflyRecovery >=10,MedianDifferenceInTsetSensitivity>=10)
colnames(goodTsetsIgnoringSpecificity) <-c("TsetName","Difference to random in redfly recovery at cutoff","Difference in Tset Sensitivity to random at cutoff","Difference to permuted in Tset Specificity at cutoff")
goodTsetsIgnoringSpecificity <- cbind(goodTsetsIgnoringSpecificity[1],goodTsetsIgnoringSpecificity[2],goodTsetsIgnoringSpecificity[3],goodTsetsIgnoringSpecificity[4])
kable_input4<- kable(goodTsetsIgnoringSpecificity,digits = 2,caption = "Good Training Sets Ignoring Poor Specificity both > 10")
#column_spec(kable_input4,2:10,width = "2cm")
kable_input4

Good Training Sets Ignoring Poor Specificity both > 10

TsetName	Difference to random in redfly recovery at cutoff	Difference in Tset Sensitivity to random at cutoff	Difference to permuted in Tset Specificity at cutoff
adult_muscle	38.49	28.75	11.11
adult_somatic_muscle	41.32	20.00	12.20
blastoderm.mapping1	47.60	22.50	6.62
emb-larv_fat_body	56.40	28.75	2.78
emb-larv_hindgut	42.66	38.57	2.77
embryonic_salivary	11.61	31.43	-0.29
eye.mapping1	56.40	20.00	3.18
haltere_disc	45.48	15.56	0.24
myoblast	48.23	15.56	0.94
ventral_ectoderm.mapping1	45.71	21.90	1.57

#write.table(goodTsetsIgnoringSpecificity,file="~/Box/NewTrainingSetsJan2020/R/goodTsetsIgnoringSpecificity.txt",sep = "\t")

2.3.2.3 Intermediate Tsets

These are the sets basically, that do not fall into very good or very bad training sets, mainly because they perform good in two measures but not in the third one. Not anymore , Excluded those sets which were already categorized as good.

Specific conditions being set for these Intermediate sets include:

• Difference To Random in redfly recovery must be grater than 8 and Difference to random in Tset Sensitivity must be greater than 8 but that of specificity should just be greater than 0 , OR
• Difference To Random in redfly recovery must be greater than 8 and specificity must be greater than 4 but Difference to random in Tset Sensitivity should just be greater than 0 , OR
• Difference To random in Tset Sensitivity and specificity must be greater than 8 and 4 respectively but Difference to random in redfly recovery should just be greater than 0.

intermediateTSets <- filter(dffinalDF,MedianDifferenceToRandomOfRedflyRecovery >=8 & MedianDifferenceInTsetSensitivity >=8 &  MedianDifferenceToPermutedInTsetSpecificity >=0 | MedianDifferenceToRandomOfRedflyRecovery >=0 & MedianDifferenceInTsetSensitivity >=8 &  MedianDifferenceToPermutedInTsetSpecificity >=4 |MedianDifferenceToRandomOfRedflyRecovery >=8 & MedianDifferenceInTsetSensitivity >=0 &  MedianDifferenceToPermutedInTsetSpecificity >=4 )
onlyintermediateTSets=data.frame()

# onlyintermediateTSetsNum <-    which(!(intermediateTSets$TsetName %in% goodTsetOverAll$TsetName))
# for (i in 1:length(onlyintermediateTSetsNum)){
# onlyintermediateTSetIter <- intermediateTSets[onlyintermediateTSetsNum[i],]
# onlyintermediateTSets <- rbind(onlyintermediateTSets,onlyintermediateTSetIter)       
# 
# }



#intermediateTSets <- onlyintermediateTSets
colnames(intermediateTSets) <-c("TsetName","Difference to random in redfly recovery at cutoff","Difference in Tset Sensitivity to random at cutoff","Difference to random in Tset Specificity at cutoff")
intermediateTSets <- cbind(intermediateTSets[1],intermediateTSets[2],intermediateTSets[3],intermediateTSets[4])
kable_input41<- kable(intermediateTSets,digits = 2,caption = "Intermediate Tsets with all 3 interm") 
#column_spec(kable_input41,2:10,width = "3cm")
kable_input41

Intermediate Tsets with all 3 interm

TsetName	Difference to random in redfly recovery at cutoff	Difference in Tset Sensitivity to random at cutoff	Difference to random in Tset Specificity at cutoff
adult_mesoderm.mapping1	47.22	6.67	7.66
adult_muscle	38.49	28.75	11.11
adult_somatic_muscle	41.32	20.00	12.20
blastoderm.mapping1	47.60	22.50	6.62
cardiac.mapping1	43.46	10.00	2.73
emb-larv_fat_body	56.40	28.75	2.78
emb-larv_hindgut	42.66	38.57	2.77
embryonic_sense_organ	3.26	21.90	4.87
embryonic_trachea	27.14	10.00	1.30
eye.mapping1	56.40	20.00	3.18
fat_body.mapping1	46.37	10.00	1.55
haltere_disc	45.48	15.56	0.24
myoblast	48.23	15.56	0.94
ventral_ectoderm.mapping1	45.71	21.90	1.57

#write.table(intermediateTSets,file="~/Box/NewTrainingSetsJan2020/R/intermediateTSets.txt",sep = "\t")

2.3.2.4 Poor Tsets

Conditions being set for these poor sets include. If a set is poor in two categories then its a poor set:

• Difference To Random in redfly recovery is less than or equalt to 8 and Difference To Random in Tset Sensitivity is also less than or equalt to 8 OR ,
• Difference To Random in Tset Sensitivity is less than or equalt to 8 and Difference To Permuted in Tset Specificity is also less than or equalt to 4 OR,
• Difference Random in redfly recovery is less than or equalt to 8 and Difference To Permuted in Tset Specificity is also less than or equalt to 4 ,

#poor Tsets

poorTsetOverAll<- filter(dffinalDF, MedianDifferenceToRandomOfRedflyRecovery <=8 & MedianDifferenceInTsetSensitivity <= 8 | MedianDifferenceInTsetSensitivity <= 8 & MedianDifferenceToPermutedInTsetSpecificity<=4 |MedianDifferenceToPermutedInTsetSpecificity<=4 & MedianDifferenceToRandomOfRedflyRecovery <=8 )

colnames(poorTsetOverAll) <-c("TsetName","Difference to random in redfly recovery at cutoff","Difference in Tset Sensitivity to random at cutoff","Difference to random in Tset Specificity at cutoff")

poorTsetOverAll<- cbind(poorTsetOverAll[1],poorTsetOverAll[2],poorTsetOverAll[3],poorTsetOverAll[4])
kable_input6<- kable(poorTsetOverAll,digits = 2,caption = "Overall Poor Training Sets with any one below 0")
#column_spec(kable_input6,2:10,width = "2cm")
kable_input6

Overall Poor Training Sets with any one below 0

TsetName	Difference to random in redfly recovery at cutoff	Difference in Tset Sensitivity to random at cutoff	Difference to random in Tset Specificity at cutoff
adult_brain	14.42	-36.10	0.81
adult_circulatory	40.87	-5.00	0.00
adult_cns	16.33	-36.34	0.89
adult_foregut	47.22	0.00	-0.48
adult_midgut	45.39	-3.16	-0.15
adult_nervous	18.42	-37.03	0.91
adult_pns	38.41	4.00	3.49
adult_sense_organ	25.33	-12.92	1.15
amnioserosa.mapping1	32.63	-0.53	3.02
antenna	26.91	-37.50	0.42
antennal_disc	9.44	-6.67	0.43
antennal_lobe	31.54	1.94	0.43
disc.mapping1	25.27	-17.27	0.75
disc.mapping2	35.75	-4.86	0.37
dorsal_ectoderm.mapping1	36.72	6.87	1.58
ectoderm.mapping1	29.23	-21.03	1.13
ectoderm.mapping2	29.95	-31.11	0.62
emb-larv_circulatory_system	47.49	4.00	0.94
emb-larv_excretory	46.89	-4.29	0.45
emb-larv_foregut	30.50	-1.11	1.27
emb-larv_visceral	32.35	4.44	0.23
emb-larval_cns	2.78	-13.21	-0.06
emb-larval_mushroombody	20.55	-5.00	1.37
emb-larval_neuron	32.79	-35.89	0.05
emb-larval_opticlobe	36.67	-2.07	1.27
embryonic_epidermis	45.67	6.67	2.80
embryonic_midgut	36.11	1.18	1.34
embryonic_muscle	49.19	5.16	2.92
embryonic_pns	32.18	-11.43	2.28
embryonic_somatic_muscle	34.78	8.00	2.61
embryonic_ventral_nervous_system	33.67	-25.29	0.60
endoderm.mapping1	35.43	-22.69	0.53
eye_disc	29.63	-5.71	1.72
eye-antennal_disc	33.94	-8.89	1.88
eye.mapping2	41.34	-24.38	0.87
female_gonad.mapping1	42.36	-12.41	1.87
female_reproductive	25.26	-4.00	1.56
genital_disc	30.98	-7.86	2.06
glia	26.86	-15.76	0.86
glia.mapping1	31.62	-2.50	1.94
glia.mapping2	27.32	-4.29	1.78
gonad	38.87	-34.74	2.57
imaginal_disc	26.23	-23.10	0.95
leg	27.90	-7.50	0.44
leg_disc	21.94	-1.54	0.52
male_reproductive	32.16	-26.96	3.92
malpig.mapping1	51.65	-2.50	2.15
mesectoderm.mapping1	47.08	5.00	3.38
mesoderm.mapping1	39.21	6.67	1.43
mesoderm.mapping2	39.32	-36.58	1.05
neuron	31.16	-36.10	0.02
pns.mapping1	27.41	-13.58	0.66
reproductive.mapping2	32.42	-35.65	2.15
salivary.mapping1	33.61	-18.95	1.27
somatic_muscle.mapping1	35.94	-7.39	2.24
trachea.mapping1	35.75	-2.50	1.20
visceral.mapping1	37.11	-0.98	1.95
wing.mapping2	21.48	-6.67	0.56

#write.table(poorTsetOverAll,file="~/Box/NewTrainingSetsJan2020/R/poorTsetOverAll.txt",sep = "\t")

2.3.3 Top None of Amplitude curve

#random 29
top_fake <-read.table("/Users/hasibaasma/Box/NewTrainingSetsJan2020/R/random62oldRun/old/del_randomMEDIAN62sets_75times_Excel_withSpec.txt",header = T)

top_orig_newMed<- top_none_imm

s1 <- top_orig_newMed[order(top_orig_newMed$TsetName),]
s2 <- top_fake[order(top_fake$TsetName),]
subsetIMMevaluationOutputContRand1000_Actual1000_diff<- cbind.data.frame(s1$TsetName,s1$Method,s1$TsetSize,s1$SCRMs,s1$TrainingSetRecovered,s1$PercentageTrainingSetSensitivity,s1$REDflyRecovered,s1$PercentageRedflyRecovered,s1$percentageExpressionPatternRecall-s1$ExpectedpercentageExpressionPatternRecall,s2$SCRMs,s2$TrainingSetRecovered,s2$PercentageTrainingSetSensitivity,s2$REDflyRecovered,s2$PercentageRedflyRecovered,s1$PercentageTrainingSetSensitivity - s2$PercentageTrainingSetSensitivity,s1$PercentageRedflyRecovered - s2$PercentageRedflyRecovered)
colnames(subsetIMMevaluationOutputContRand1000_Actual1000_diff)[1] <-"TsetName"
colnames(subsetIMMevaluationOutputContRand1000_Actual1000_diff)[4] <-"SCRMs"
colnames(subsetIMMevaluationOutputContRand1000_Actual1000_diff)[9] <-"DifferenceBetweenActualRandomPercentageSpecificity"
colnames(subsetIMMevaluationOutputContRand1000_Actual1000_diff)[15] <- "DifferenceBetweenActualRandomTsetRecovery"
colnames(subsetIMMevaluationOutputContRand1000_Actual1000_diff)[16] <- "DifferenceBetweenActualRandomRedflyRecovery"

##checking..creating ranges of differences to write in the observations
minimum_differenceBetweenActualRandomRedflyRecovery <- list()
maximum_differenceBetweenActualRandomRedflyRecovery <- list()
median_differenceBetweenActualRandomRedflyRecovery <- list()
minimum_DifferenceBetweenActualTsetRecovery<- list()
maximum_DifferenceBetweenActualTsetRecovery<- list()
median_DifferenceBetweenActualTsetRecovery<- list()
minimum_specificity<- list()
maximum_specificity<- list()
median_specificity<- list()
for(i in unique(subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName)){
  minimum_differenceBetweenActualRandomRedflyRecovery[[i]] <- min(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomRedflyRecovery[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  median_differenceBetweenActualRandomRedflyRecovery[[i]] <- median(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomRedflyRecovery[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  maximum_differenceBetweenActualRandomRedflyRecovery[[i]] <-max(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomRedflyRecovery[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  minimum_DifferenceBetweenActualTsetRecovery[[i]]<-min(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomTsetRecovery[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  maximum_DifferenceBetweenActualTsetRecovery[[i]]<-max(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomTsetRecovery[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  median_DifferenceBetweenActualTsetRecovery[[i]]<-median(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomTsetRecovery[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  
  minimum_specificity[[i]]<-min(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomPercentageSpecificity[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  maximum_specificity[[i]]<-max(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomPercentageSpecificity[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  median_specificity[[i]]<-median(subsetIMMevaluationOutputContRand1000_Actual1000_diff$DifferenceBetweenActualRandomPercentageSpecificity[subsetIMMevaluationOutputContRand1000_Actual1000_diff$TsetName==i])*100
  
}
library(qdapTools)
minred <- list2df(minimum_differenceBetweenActualRandomRedflyRecovery,"minimum difference to random in redfly recovery","TsetName")
medred <-list2df(median_differenceBetweenActualRandomRedflyRecovery,"median difference to random in redfly recovery","TsetName")
maxred<-list2df(maximum_differenceBetweenActualRandomRedflyRecovery,"maximum difference to random in redfly recovery","TsetName")
minTset<-list2df(minimum_DifferenceBetweenActualTsetRecovery,"minimum difference to random in Tset sensitivity","TsetName")
maxTset<- list2df(maximum_DifferenceBetweenActualTsetRecovery,"maximum difference to random in Tset sensitivity","TsetName")
medTset<-list2df(median_DifferenceBetweenActualTsetRecovery,"median difference to random in Tset sensitivity","TsetName")
minSpec<-list2df(minimum_specificity,"minimum difference to random in specificity","TsetName")
maxSpec<-list2df(maximum_specificity,"maximum difference to random in specificity","TsetName")
medSpec <-list2df(median_specificity,"median difference to random in specificity","TsetName")

df1<- merge(medred,medTset,by="TsetName")
dffinal<-merge(df1,medSpec,by="TsetName")
#dffinal<-merge(df7,by="TsetName")




#data.frame(unclass(table(dffinal)))
#table(dffinal)
write.table(dffinal,file="~/Box/Old_and_newTsets_3postProcMethods_3categories/newTsets/finaltable_new74sets_noneAmplitudeCurve.txt",sep = "\t")

library(knitr)
library(kableExtra)
library(dplyr)

2.3.3.1 Good Tsets

Conditions being set for these good Tsets include:

• Difference To Random in Redfly Recovery must be greater than or equal to 8percent at cutoff,
• Difference to Random in Tset Sensitivity must be greater than or equalt to 8percent,
• Difference to Expected in Specificity must be greater than or equalt to 4percent

finaltable<- list()
# 
#setting up conditions Good Tsets OverAll
dffinalDF <- as.data.frame(dffinal)


colnames(dffinalDF) <-c("TsetName","MedianDifferenceToRandomOfRedflyRecovery","MedianDifferenceInTsetSensitivity","MedianDifferenceToPermutedInTsetSpecificity")


goodTsetOverAll<- filter(dffinalDF,MedianDifferenceToRandomOfRedflyRecovery >=8,  MedianDifferenceInTsetSensitivity >=8,MedianDifferenceToPermutedInTsetSpecificity >=4)



goodTsetsOverAll <- list()
colnames(goodTsetOverAll) <-c("TsetName","Difference to random in redfly recovery at cutoff","Difference in Tset Sensitivity to random at cutoff","Difference to permuted in Tset Specificity at cutoff")

goodTsetOverAll<- cbind(goodTsetOverAll[1],goodTsetOverAll[2],goodTsetOverAll[3],goodTsetOverAll[4])
kable_input3<- kable(goodTsetOverAll,digits = 2,caption = "Overall Good Training Sets with TS and RR > 8 and SP > 4") 
#column_spec(kable_input3,2:10,width = "2cm")
kable_input3

Overall Good Training Sets with TS and RR > 8 and SP > 4

TsetName	Difference to random in redfly recovery at cutoff	Difference in Tset Sensitivity to random at cutoff	Difference to permuted in Tset Specificity at cutoff
adult_mesoderm.mapping1	27.08	26.67	10.65
adult_muscle	23.12	35.00	10.89
adult_somatic_muscle	24.25	20.00	11.71
amnioserosa.mapping1	27.46	10.00	5.78
blastoderm.mapping1	17.27	35.00	14.10
emb-larv_excretory	26.98	17.14	6.25
emb-larv_fat_body	9.51	41.25	5.00
emb-larv_hindgut	19.44	45.71	5.11
embryonic_muscle	26.20	27.74	5.00
embryonic_pns	20.71	17.14	5.28
embryonic_somatic_muscle	23.45	36.00	4.67
embryonic_trachea	17.90	22.50	4.07
eye_disc	16.49	8.57	4.40
eye.mapping1	23.31	30.00	4.11
malpig.mapping1	27.28	10.00	6.92
mesectoderm.mapping1	20.23	30.00	6.47
mesoderm.mapping1	22.59	24.44	4.96
somatic_muscle.mapping1	25.21	10.00	6.24
visceral.mapping1	24.61	13.66	5.58

#write.table(goodTsetOverAll,file="~/Box/NewTrainingSetsJan2020/R/goodTsetOverAll.txt",sep = "\t")

2.3.3.2 Good Tsets Ignoring Specificity

Conditions being set for these good Tsets(if we ignore specificity) include:

• Difference To Random in redfly recovery must be greater than or equal to 10percent at cutoff,
• Difference In Tset Sensitivity to random must be greater than or equalt to 10.

goodTsetsIgnoringSpecificity <- filter(dffinalDF,MedianDifferenceToRandomOfRedflyRecovery >=10,MedianDifferenceInTsetSensitivity>=10)
colnames(goodTsetsIgnoringSpecificity) <-c("TsetName","Difference to random in redfly recovery at cutoff","Difference in Tset Sensitivity to random at cutoff","Difference to permuted in Tset Specificity at cutoff")
goodTsetsIgnoringSpecificity <- cbind(goodTsetsIgnoringSpecificity[1],goodTsetsIgnoringSpecificity[2],goodTsetsIgnoringSpecificity[3],goodTsetsIgnoringSpecificity[4])
kable_input4<- kable(goodTsetsIgnoringSpecificity,digits = 2,caption = "Good Training Sets Ignoring Poor Specificity both > 10")
#column_spec(kable_input4,2:10,width = "2cm")
kable_input4

Good Training Sets Ignoring Poor Specificity both > 10

TsetName	Difference to random in redfly recovery at cutoff	Difference in Tset Sensitivity to random at cutoff	Difference to permuted in Tset Specificity at cutoff
adult_mesoderm.mapping1	27.08	26.67	10.65
adult_muscle	23.12	35.00	10.89
adult_somatic_muscle	24.25	20.00	11.71
antennal_lobe	16.42	11.61	1.43
blastoderm.mapping1	17.27	35.00	14.10
cardiac.mapping1	25.05	13.57	3.97
disc.mapping2	19.82	11.35	3.00
dorsal_ectoderm.mapping1	18.97	19.37	3.65
emb-larv_circulatory_system	23.35	16.00	3.37
emb-larv_excretory	26.98	17.14	6.25
emb-larv_hindgut	19.44	45.71	5.11
emb-larv_visceral	21.06	15.56	0.91
embryonic_midgut	27.57	18.82	2.62
embryonic_muscle	26.20	27.74	5.00
embryonic_pns	20.71	17.14	5.28
embryonic_somatic_muscle	23.45	36.00	4.67
embryonic_trachea	17.90	22.50	4.07
eye.mapping1	23.31	30.00	4.11
female_reproductive	11.33	12.00	3.33
haltere_disc	23.55	15.56	1.51
leg_disc	14.53	13.85	2.31
mesectoderm.mapping1	20.23	30.00	6.47
mesoderm.mapping1	22.59	24.44	4.96
myoblast	15.69	26.67	1.15
ventral_ectoderm.mapping1	11.45	21.90	2.06
visceral.mapping1	24.61	13.66	5.58

#write.table(goodTsetsIgnoringSpecificity,file="~/Box/NewTrainingSetsJan2020/R/goodTsetsIgnoringSpecificity.txt",sep = "\t")

2.3.3.3 Intermediate Tsets

These are the sets basically, that do not fall into very good or very bad training sets, mainly because they perform good in two measures but not in the third one. Not anymore , Excluded those sets which were already categorized as good.

Specific conditions being set for these Intermediate sets include:

• Difference To Random in redfly recovery must be grater than 8 and Difference to random in Tset Sensitivity must be greater than 8 but that of specificity should just be greater than 0 , OR
• Difference To Random in redfly recovery must be greater than 8 and specificity must be greater than 4 but Difference to random in Tset Sensitivity should just be greater than 0 , OR
• Difference To random in Tset Sensitivity and specificity must be greater than 8 and 4 respectively but Difference to random in redfly recovery should just be greater than 0.

intermediateTSets <- filter(dffinalDF,MedianDifferenceToRandomOfRedflyRecovery >=8 & MedianDifferenceInTsetSensitivity >=8 &  MedianDifferenceToPermutedInTsetSpecificity >=0 | MedianDifferenceToRandomOfRedflyRecovery >=0 & MedianDifferenceInTsetSensitivity >=8 &  MedianDifferenceToPermutedInTsetSpecificity >=4 |MedianDifferenceToRandomOfRedflyRecovery >=8 & MedianDifferenceInTsetSensitivity >=0 &  MedianDifferenceToPermutedInTsetSpecificity >=4 )
onlyintermediateTSets=data.frame()

# onlyintermediateTSetsNum <-    which(!(intermediateTSets$TsetName %in% goodTsetOverAll$TsetName))
# for (i in 1:length(onlyintermediateTSetsNum)){
# onlyintermediateTSetIter <- intermediateTSets[onlyintermediateTSetsNum[i],]
# onlyintermediateTSets <- rbind(onlyintermediateTSets,onlyintermediateTSetIter)       
# 
# }



#intermediateTSets <- onlyintermediateTSets
colnames(intermediateTSets) <-c("TsetName","Difference to random in redfly recovery at cutoff","Difference in Tset Sensitivity to random at cutoff","Difference to random in Tset Specificity at cutoff")
intermediateTSets <- cbind(intermediateTSets[1],intermediateTSets[2],intermediateTSets[3],intermediateTSets[4])
kable_input41<- kable(intermediateTSets,digits = 2,caption = "Intermediate Tsets with all 3 interm") 
#column_spec(kable_input41,2:10,width = "3cm")
kable_input41

Intermediate Tsets with all 3 interm

TsetName	Difference to random in redfly recovery at cutoff	Difference in Tset Sensitivity to random at cutoff	Difference to random in Tset Specificity at cutoff
adult_mesoderm.mapping1	27.08	26.67	10.65
adult_muscle	23.12	35.00	10.89
adult_somatic_muscle	24.25	20.00	11.71
amnioserosa.mapping1	27.46	10.00	5.78
antennal_lobe	16.42	11.61	1.43
blastoderm.mapping1	17.27	35.00	14.10
cardiac.mapping1	25.05	13.57	3.97
disc.mapping2	19.82	11.35	3.00
dorsal_ectoderm.mapping1	18.97	19.37	3.65
emb-larv_circulatory_system	23.35	16.00	3.37
emb-larv_excretory	26.98	17.14	6.25
emb-larv_fat_body	9.51	41.25	5.00
emb-larv_foregut	21.61	7.22	6.17
emb-larv_hindgut	19.44	45.71	5.11
emb-larv_visceral	21.06	15.56	0.91
emb-larval_opticlobe	24.16	8.28	3.04
embryonic_epidermis	32.66	6.67	6.50
embryonic_midgut	27.57	18.82	2.62
embryonic_muscle	26.20	27.74	5.00
embryonic_pns	20.71	17.14	5.28
embryonic_sense_organ	0.44	26.67	9.61
embryonic_somatic_muscle	23.45	36.00	4.67
embryonic_trachea	17.90	22.50	4.07
eye_disc	16.49	8.57	4.40
eye-antennal_disc	18.65	2.22	4.70
eye.mapping1	23.31	30.00	4.11
fat_body.mapping1	7.47	35.00	4.31
female_gonad.mapping1	14.79	1.38	4.13
female_reproductive	11.33	12.00	3.33
haltere_disc	23.55	15.56	1.51
leg_disc	14.53	13.85	2.31
malpig.mapping1	27.28	10.00	6.92
mesectoderm.mapping1	20.23	30.00	6.47
mesoderm.mapping1	22.59	24.44	4.96
myoblast	15.69	26.67	1.15
pns.mapping1	20.50	9.06	3.17
salivary.mapping1	23.04	2.11	7.51
somatic_muscle.mapping1	25.21	10.00	6.24
ventral_ectoderm.mapping1	11.45	21.90	2.06
visceral.mapping1	24.61	13.66	5.58
wing.mapping2	18.33	8.48	3.40

#write.table(intermediateTSets,file="~/Box/NewTrainingSetsJan2020/R/intermediateTSets.txt",sep = "\t")

2.3.3.4 Poor Tsets

Conditions being set for these poor sets include. If a set is poor in two categories then its a poor set:

• Difference To Random in redfly recovery is less than or equalt to 8 and Difference To Random in Tset Sensitivity is also less than or equalt to 8 OR ,
• Difference To Random in Tset Sensitivity is less than or equalt to 8 and Difference To Permuted in Tset Specificity is also less than or equalt to 4 OR,
• Difference Random in redfly recovery is less than or equalt to 8 and Difference To Permuted in Tset Specificity is also less than or equalt to 4 ,

#poor Tsets

poorTsetOverAll<- filter(dffinalDF, MedianDifferenceToRandomOfRedflyRecovery <=8 & MedianDifferenceInTsetSensitivity <= 8 | MedianDifferenceInTsetSensitivity <= 8 & MedianDifferenceToPermutedInTsetSpecificity<=4 |MedianDifferenceToPermutedInTsetSpecificity<=4 & MedianDifferenceToRandomOfRedflyRecovery <=8 )

colnames(poorTsetOverAll) <-c("TsetName","Difference to random in redfly recovery at cutoff","Difference in Tset Sensitivity to random at cutoff","Difference to random in Tset Specificity at cutoff")

poorTsetOverAll<- cbind(poorTsetOverAll[1],poorTsetOverAll[2],poorTsetOverAll[3],poorTsetOverAll[4])
kable_input6<- kable(poorTsetOverAll,digits = 2,caption = "Overall Poor Training Sets with any one below 0")
#column_spec(kable_input6,2:10,width = "2cm")
kable_input6

Overall Poor Training Sets with any one below 0

TsetName	Difference to random in redfly recovery at cutoff	Difference in Tset Sensitivity to random at cutoff	Difference to random in Tset Specificity at cutoff
adult_brain	9.48	-25.71	2.06
adult_circulatory	21.46	5.00	3.03
adult_cns	10.81	-27.80	1.69
adult_midgut	21.91	7.37	-1.54
adult_nervous	11.96	-32.08	2.10
adult_sense_organ	6.89	10.00	3.85
antenna	17.32	-23.75	1.58
antennal_disc	6.07	2.86	0.79
disc.mapping1	19.35	-5.15	3.39
ectoderm.mapping1	21.87	-7.24	1.60
ectoderm.mapping2	23.02	-15.56	2.19
emb-larval_cns	-2.90	1.07	0.37
emb-larval_mushroombody	10.33	5.00	3.51
emb-larval_neuron	17.51	-29.04	0.37
embryonic_ventral_nervous_system	19.30	-12.06	2.17
endoderm.mapping1	24.43	-3.46	3.04
eye.mapping2	26.70	-10.31	3.85
genital_disc	22.82	6.43	1.42
glia	19.27	2.42	1.36
glia.mapping1	17.07	5.00	3.61
glia.mapping2	17.42	2.86	2.65
imaginal_disc	13.54	-16.06	3.27
leg	21.55	5.00	3.95
mesoderm.mapping2	28.51	-29.04	3.76
neuron	16.04	-32.21	0.43
trachea.mapping1	18.05	5.83	0.74

#write.table(poorTsetOverAll,file="~/Box/NewTrainingSetsJan2020/R/poorTsetOverAll.txt",sep = "\t")